272 80 17MB
English Pages 431 Year 2011
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved. Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved. Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
MATERIALS SCIENCE AND TECHNOLOGIES
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
COMPOSITE MATERIALS IN ENGINEERING STRUCTURES
No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in renderingStructures, legal, medical or anyPublishers, other professional services. Composite Materials in Engineering Nova Science Incorporated, 2011. ProQuest Ebook Central,
MATERIALS SCIENCE AND TECHNOLOGIES Additional books in this series can be found on Nova’s website under the Series tab.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Additional E-books in this series can be found on Nova’s website under the E-book tab.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
MATERIALS SCIENCE AND TECHNOLOGIES
COMPOSITE MATERIALS IN ENGINEERING STRUCTURES
JENNIFER M. DAVIS
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
EDITOR
New York Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011 by Nova Science Publishers, Inc. All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Any parts of this book based on government reports are so indicated and copyright is claimed for those parts to the extent applicable to compilations of such works.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS. Additional color graphics may be available in the e-book version of this book. LIBRARY OF CONGRESS CATALOGING-IN-PUBLICATION DATA Composite materials in engineering structures / editor, Jennifer M. Davis. p. cm. Includes bibliographical references. ISBN (H%RRN) 1. Composite materials. 2. Engineering design. I. Davis, Jennifer M. TA418.9.C6C583 2010 624.1'8--dc22 2010026088
Published by Nova Science Publishers, Inc. † New York
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
CONTENTS
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Preface
vii
Chapter 1
Effects of Thermo-Oxidation on Composite Materials and Structures at High Temperatures Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Chapter 2
Damping in Composite Materials and Structures Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani and Abderrahim El Mahi
Chapter 3
Mechanical States Induced by Moisture Diffusion in Organic Matrix Composites: Coupled Scale Transition Models F. Jacquemin and S. Fréour
137
Chapter 4
Fatigue and Fracture of Short Fibre Composites Exposed to Extreme Temperatures B.G. Prusty and J. Sul
191
Chapter 5
Fatigue of Polymer Matrix Composites at Elevated Temperatures - A Review John Montesano, Zouheir Fawaz, Kamran Behdinan and Cheung Poon
229
Chapter 6
The Closed Form Solutions of Infinitesimal and Finite Deformation of 2-D Laminated Curved Beams of Variable Curvatures K.C. Lin and C.M. Hsieh
253
Chapter 7
Development and Application of Fibre-Reinforced Metal Laminates in Aerospace Structures P. Terry Crouch and Y.X. Zhang
293
Chapter 8
Critical Aeroelastic Behaviour of Slender Composite Wings in an Incompressible Flow Enrico Cestino and Giacomo Frulla
313
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
1
53
vi
Contents
Chapter 9
Exact Solution for the Postbuckling of Composite Beams Samir A. Emam
341
Chapter 10
Buckling Behaviors of Elastic Functionally Graded Cylindrical Shells Huaiwei Huang and Qiang Han
373
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Index
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
409
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
PREFACE Composite materials such as fiber-reinforced composites, aggregate composites, and natural fiber reinforced composites have been used widely in engineering structures in various industries. Composite laminates, especially fiber reinforced metal laminates (FRMLs) have been used extensively in aerospace structures. Composite laminates are materials that involve some combination on a macroscopic scale of two or more different primary structural engineering constituents such as polymers, metals, ceramics and glasses. This book presents current research from across the globe in the study of composite materials, including the effects of thermo-oxidation on composite materials and structures at high temperatures; damping in composite materials; fatigue and fracture of short fiber composites; and solutions for postbuckling of composite beams. Chapter 1 reviews some research activity carried out since several years by the members of the Physics and Mechanics Department – Insitut Pprime – ENSMA concerning the effects of thermo-oxidation in composite materials and structures at high temperatures (T > 120°C) and aims at giving a quite comprehensive understanding of ageing phenomena occurring under thermo-oxidative environments. Thermo-oxidation is a coupled oxygen reaction-diffusion phenomenon occurring in polymer matrices at high temperatures which may lead to irreversible shrinkage strains, local mechanical properties changes, fibre-matrix debonding and matrix cracking onset close to the external surfaces of composite materials and structures exposed to air or oxygen rich environments, reducing their durability performances. In the present chapter first thermo-oxidation phenomena are introduced and a comprehensive literature review is given. Confocal interferometric spectroscopy methods are introduced as a tool to measure local thermo-oxidation induced shrinkage strains and deformations at the exposed edges of composite samples. A multi-physics unified model approach based on the thermodynamics of irreversible processes is then presented. The model is able to put in evidence chemo-mechanics couplings and shows how the oxygen reaction-diffusion within the polymer may be influenced by the mechanical variables. A classical mechanistic model for oxygen reaction-diffusion can be also recovered by introducing simplifying hypotheses. The model is then identified and validated through different experimental tests and – once validated – satisfactorily employed for the simulation of thermo-oxidation induced local strains and stresses in composites under several different environmental conditions. Such
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
viii
Jennifer M. Davis
simulations help identifying the critical conditions for thermo-oxidation induced damage onset and propagation. The possibility to accelerate thermo-oxidation ageing phenomena through increasing oxygen pressure is investigated and discussed both experimentally and theoretically. The purpose of Chapter 2 is to report an extended synthesis of the recent developments on the evaluation of the damping of laminates and sandwich materials. Modelling of damping as well as experimental investigation will be considered. The different concepts introduced will be last applied to the analysis of the dynamic response of a simple shape damped composite structure. Composite structures are often submitted to hygroscopic loads during their service life. Moisture uptake generates multi-scale internal stresses, the knowledge of which, granted by dedicated scale transition approaches, is precious for sizing mechanical part or predicting their durability. Experiments report that the diffusion properties of penetrant-organic matrix composite systems may continuously change during the diffusion process, due to the evolution of the internal strains experienced by the polymer matrix. On the one hand, both the diffusion coefficient and the maximum moisture absorption capacity, i.e. the main penetrant transport factors, are affected by the distribution of the local strains within the composite structure. On the other hand, accounting for strain dependent diffusion parameters change the moisture content profiles, which affect the mechanical states distribution itself. Consequently, a strong two-ways hygro-mechanical coupling occurs in organic matrix composites. The literature also reports that the effective stiffness tensor of composite plies is directly linked to their moisture content. Actually, the main parameters controlling the diffusion process remain unchanged. Thus, only the time- and depth- dependent mechanical states are affected. Consequently, this effect, independently handled, constitutes a single-way hygromechanical coupling by comparison with the above described phenomenon. This work investigates the consequences of accounting for such coupling in the modelling of the hygromechanical behaviour of composites structures through scale transitions approaches. The first part of Chapter 3 deals with the effects related to the moisture content dependent evolution of the hygro-elastic properties of composite plies on the in-depth stress states predicted during the transient stage of the diffusion process. The numerical simulations show that accounting (or not) for the softening of the materials properties occurring in practice, yields significant discrepancies of the predicted multi-scale stress states. In a second part, the free-volume theory is introduced in the multi-scale hygro-mechanical model in order to achieve the coupling between the mechanical states experienced by the organic matrix and its diffusion controlling parameters. Various numerical practical cases are considered: the effect of the internal swelling strains on the time- and depth-dependent diffusion coefficient, maximum moisture absorption capacity, moisture content and internal stresses states are studied and discussed. Homogenization relations are required for estimating macroscopic diffusion coefficients from those of the plies constituents. In the third part, effective diffusivities of composite plies are estimated from the solving of unit cell problems over representative volume elements submitted to macroscopic moisture gradients when accounting for the resulting mechanical states profiles. As discussed in Chapter 4, fibre-reinforced composites have been used for more than 50 years and are still being evolving in terms of material integrity, manufacturing process and its
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Preface
ix
performance under adverse conditions. The advent of graphite fibres from polyacrylonitrile organic polymer has resulted in a high performance material, namely carbon based composites, performing better in every respect than glass fibre-reinforced plastic (GFRP). However, glass fibres are still in high demand for wide applications, where the cost takes precedence over performance. Owing to its quasi-isotropic properties, randomly orientated short fibre reinforced composites, particularly chopped strand mat (CSM) and sheet moulding compound, are playing a critical role in boat building industry and automotive industry, respectively. As structural performance of composite material is being improved, GFRPs are expected to replace metals in more harsh applications, in which high cyclic loadings and elevated temperatures are applied. Furthermore, heat deflection temperature of common thermosetting resin is in the range from 65ºC to 85ºC under applied stress of 1.8MPa. The thermal effects on short-fibre thermosetting composites have not been flourishingly investigated. Fatigue prediction of mechanical structure is not only critical at the design stage, but is much more critical for the maintenance strategy. The fatigue, fracture and durability of GFRP-CSM are complex issues because of so many variables contributing to thermal and mechanical damages. Despite a number of approaches to modeling fatigue damage of GFRP using phenomenological methodologies based on the strength and stiffness degradation, or physical modelling based on micro-mechanics, their performance under adverse thermo-mechanical loading has not been fully understood to benefit the end users. In recent years, advanced composite materials have been frequently selected for aerospace applications due to their light weight and high strength. Polymer matrix composite (PMC) materials have also been increasingly considered for use in elevated temperature applications, such as supersonic vehicle airframes and propulsion system components. A new generation of high glass-transition temperature polymers has enabled this development to materialize. Clearly, there is a requirement to better understand the mechanical behaviour of this class of composite materials in order to achieve widespread acceptance in practical applications. More specifically, an improved understanding of the behaviour of PMC materials when subjected to elevated temperature cyclic loading is warranted. Chapter 5 contains a comprehensive review of the experimental and numerical studies conducted on various PMC materials subjected to elevated temperature fatigue loading. Experimental investigations typically focus on observing damage phenomenon and time-dependent material behaviour exhibited during elevated temperature testing, whereas insufficient fatigue test data is found in the literature. This is mainly due to the long-term high temperature limitations of most conventional PMC materials and of the experimental equipment. Moreover, it has been found that few fatigue models have been developed that are suitable for damage progression simulations of PMC materials during elevated temperature fatigue loading. Although this review is not exhaustive, the noteworthy results and trends of the most important studies are presented, as well as their apparent shortcomings. Lastly, recommendations for future studies are addressed and the focus of current research efforts is outlined. The analytical solutions of infinitesimal deformation and finite deformation for in-plane slender laminated curved beams of variable curvatures are developed in Chapter 6. The effects of aspect ratio, thickness ratio, stacking sequence and material orthotropic ratios on the laminated curved beams or rings are presented. By introducing the variables of curvature and angle of tangent slope, the governing equations for the infinitesimal deformation analysis are expressed in terms of un-deformed
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
x
Jennifer M. Davis
configuration. All the quantities of axial force, shear force, moment, and displacements are decoupled and expressed in terms of tangent angle. The first and second moments of arc length with respect to horizontal and vertical axes of curved beams are defined as fundamental geometric properties. The analytical solutions of circular, elliptical, parabola, cantenary, cycloid, spiral curved beams under various loading are demonstrated. The circular ring under point load and distributed load is presented as well. The analytical solutions are consistent with published results. The governing equations for finite deformation analysis are expressed in terms of deformed configuration. All the quantities are formulated as functions of angle of tangent slope in deformed state. The analytical solutions of laminated circular curved beam under pure bending are presented. The results show that the circular curved beam remains as a circular curved beam during deformation. Development of fiber reinforced metal laminates (FRMLs) and their applications in aerospace structures are reviewed in Chapter 7, especially for Glass-reinforced Aluminium Laminates (GLARE) currently used extensively in aerospace industry, Central reinforced Aluminums (CentrAL) and Hybrid Titanium Composite Laminates (HTCL), which show strong signs to become dominating FRMLs for aerospace applications in the future. Nonlinear finite element analyses are carried out using the commerical finite element software ANSYS to investigate the structural behaviour of these three FRMLs. The effects of specific parameters such as volumetric fibre content, matrix thickness, lay-up configuration, and fibre orientation on deflection and stress behaviour of GLARE, CentrAL and HTCL are also investigated in this chapter. The different responses of the structural behaviour from the different FRMLs are compared. As explained in Chapter 8, the design of highly flexible aircraft, such as high-altitude long endurance (HALE) configurations, must include phenomena that are not usually considered in traditional aircraft design. Wing flexibility, coupled with long wing span can lead to large deflections during normal flight operation with aeroelastic instabilities quite different from their rigid counterparts. A proper beam model, capable of describing the structural flight deflections, should be adopted. It includes the evaluation of the equivalent stiffness both in the case of isotropic configuration and in simple/thin-walled laminated sections emphasizing different coupling effects. Consequently, the flutter analysis has to be performed considering the deflected state as a reference point. The resulting equations are derived by the extended Hamilton's principle and are valid to second order for long, slender, composite beams undergoing moderate to large displacements. The structural model has been coupled with an unsteady aerodynamic model for an incompressible flow field, based on the Wagner aerodynamic indicial function, in order to obtain a nonlinear aeroelastic model. Using Galerkin's method and a mode summation technique, the governing equations will be solved by introducing a simple numerical method that enables one to expedite the calculation process during the preliminary design phase. In order to assess the accuracy of the prediction, the results obtained in a test case are compared with a FEM model showing a good correlation. The effect of typical parameters on critical boundaries, including stiffness ratios, ply layup, deflection amplitude, as well as the wing aspect ratio, are investigated. Analytical/Experimental comparisons are presented both in the linear case and in the non-linear derivation. A test model identification procedure is also reported, based on similarity theory, for the development of a wind-tunnel component suitable for experimental test campaign.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Preface
xi
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Chapter 9 has two main parts: the first part deals with composite beams without imperfection, and the second part is about composite beams exhibiting a geometric imperfection. It will be shown that the lay up of the composite laminate and the initial imperfection can be used as two control parameters to enhance the beam’s response. Functionally graded materials (FGMs) are microscopically inhomogeneous in which the mechanical properties vary smoothly and continuously from one surface to the other. Recent years have witnessed extensive investigations on this new class of materials due to their high performances on heat resisting and crack preventing. In stability analyses of FGM structures, buckling of FGM cylindrical shells has always been concerned. Chapter 10 systematically illustrates buckling and postbuckling behaviors of FGM cylindrical shells under combined loads. Firstly, linear buckling of FGM cylindrical shells is investigate by using the Stein prebuckling consistent theory which takes into account the effect of shell’s prebuckling deflection. Linear results are verified theoretically. However, there is generally a huge difference for buckling critical load between linear prediction and experiments of homogeneous cylindrical shell structures. To reveal this difference in the FGM case, the geometrical nonlinearity of FGM cylindrical shells is considered subsequently. It shows clearly that the theoretically-predicted linear and nonlinear buckling critical loads give respectively the upper and the lower limit of the experimental one. Meanwhile, postbuckling behaviors of FGM cylindrical shells are studied as well. Because FGMs usually serve in thermal environment, thermal effects on buckling of FGM cylindrical shells are also discussed. Besides, numerical results show the effects of the inhomogeneous parameter of FGMs, the dimensional parameter and so on.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved. Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
In: Composite Materials in Engineering Structures Editor: Jennifer M. Davis
ISBN: 978-1-61728-857-9 © 2011 Nova Science Publishers, Inc.
Chapter 1
EFFECTS OF THERMO-OXIDATION ON COMPOSITE MATERIALS AND STRUCTURES AT HIGH TEMPERATURES Marco Gigliotti*, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot Institut Pprime, CNRS – ENSMA – Université de Poitiers, Département Physique et Mécanique des Matériaux, ENSMA Téleport 2 – 1, Avenue Clement Ader, BP 40109 - F86961 Futuroscope Chasseneuil Cedex, France
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
ABSTRACT The present chapter reviews some research activity carried out since several years by the members of the Physics and Mechanics Department – Insitut Pprime – ENSMA concerning the effects of thermo-oxidation in composite materials and structures at high temperatures (T > 120°C) and aims at giving a quite comprehensive understanding of ageing phenomena occurring under thermo-oxidative environments. Thermo-oxidation is a coupled oxygen reaction-diffusion phenomenon occurring in polymer matrices at high temperatures which may lead to irreversible shrinkage strains, local mechanical properties changes, fibre-matrix debonding and matrix cracking onset close to the external surfaces of composite materials and structures exposed to air or oxygen rich environments, reducing their durability performances. In the present chapter first thermo-oxidation phenomena are introduced and a comprehensive literature review is given. Confocal interferometric spectroscopy methods are introduced as a tool to measure local thermo-oxidation induced shrinkage strains and deformations at the exposed edges of composite samples. A multi-physics unified model approach based on the thermodynamics of irreversible processes is then presented. The model is able to put in evidence chemo*
Corresponding author: Tel.: +33 0549 49 8340, Fax: +33 0549 49 8238 (Secretariat), E-mail: marco.gigliotti @lmpm.ensma.fr
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
2
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot mechanics couplings and shows how the oxygen reaction-diffusion within the polymer may be influenced by the mechanical variables. A classical mechanistic model for oxygen reaction-diffusion can be also recovered by introducing simplifying hypotheses. The model is then identified and validated through different experimental tests and – once validated – satisfactorily employed for the simulation of thermo-oxidation induced local strains and stresses in composites under several different environmental conditions. Such simulations help identifying the critical conditions for thermo-oxidation induced damage onset and propagation. The possibility to accelerate thermo-oxidation ageing phenomena through increasing oxygen pressure is investigated and discussed both experimentally and theoretically.
Keywords: Polymer Matrix Composites (PMCs), thermo-oxidation, multi-physics couplings, multiscale modeling, viscoelasticity, ultra-micro indentation (UMI), confocal interferometric microscopy (CIM), scanning electronic microscopy (SEM).
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
INTRODUCTION Starting from the second half of the 20th century, composite materials have been massively employed for the realisation of aerospace structures. Glass fibre / organic resin carbon fibre / epoxy resin composites (Polymer Matrix Composites, PMCs) are used for helicopters and civil aircraft applications since the 1950s and today this trend does not stop growing. In fact, these materials reduce the cost of structures, by reducing the weight, the number of manufactured parts and the maintenance during service. The weight gain which can be obtained by the employment of composite parts is about 20%, and may even reach 55% in some cases (helicopters structures made of aramid / epoxy composites). In addition to this, the cost of the raw material has progressively decreased over time (the cost of a high-strength carbon fiber was 300 € / kg in the 1970s and around 30 € / kg today) making composite materials particularly attractive to designers and industrial producers. It is superfluous to say that material saving and reduction of manufactured parts promote environmental benefits, besides economical. Composite materials exhibit very high specific stiffness and strength values and good fatigue performance. Moreover their properties can be tailored to reach some specific targets. The optimal employment of such materials implies that new routes are currently explored, including the use of composites for structures subjected to severe and aggressive environmental solicitations. This is the case for instance of ‘hot’ aeronautical structures, structures employed for supersonic flights or for turbo engines, where the temperatures can be as high as 180°C (depending on the application) and where the presence of oxygen in the environment induces accelerated resin polymer chemical ageing (thermo-oxidation in particular) leading to dramatic decrease of the part’s durability. Chemical ageing can be seen as a set of irreversible changes occurring in a polymer material due to chain scission, reaction-diffusion phenomena, crosslinking, hydrolysis: it often takes place together (or in competition) with physical ageing - the reversible polymer evolution towards an thermodynamic equilibrium state - and both phenomena are active on the long time scale. Though the fibres are not particularly sensitive to such phenomena, the
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Effects of Thermo-Oxidation on Composite Materials and Structures …
3
PMC as a whole can be consistently affected by polymer degradation; moreover - in correspondence with the three-dimensional material zone which physically exist at the frontier between fibres and polylmer matrix (interphase) – degradation phenomena cannot be clearly singled out due to the complex chemical composition of the interphase and fibrematrix interfaces can become the place of very high solicitations. These phenomena may in turn lead to fibre-matrix debonding and spontaneous cracking, which results in diffuse damage onset and propagation. Lafarie-Frenot [1] has shown that a thermo-oxidizing environment can increase consistently the density and the growth kinetics of matrix “mesocracks” (cracks at the ply scale) in the off-axis plies of PMCs samples subjected to thermal fatigue and may have an important impact on the damage tolerance performances of such materials. Today, the design of composite structures implies the use of knockdown factors to take into full account the complex effects of the degradation phenomena occurring in PMCs. Such factors are established on empirical (and often uncertain) bases without clear links to the physical phenomena occurring during material ageing; this leads – from one side - to excessively conservative design (which is far from being optimal from the economical viewpoint) and – from another side – to poor damage tolerance performances. Therefore, structures made of such materials may require massive maintenance operations. A rational approach to the damage tolerant design of long term PMC based ‘hot’ structures requires a consistent research effort in order to elucidate the basic degradation mechanisms and quantify their kinetics. Degradation phenomena of different nature may be effective at the same timescales and interact with each other giving rise to coupled effects: moreover such couplings may be apparent at different structural scales. Therefore, this behavior can be explored only by means of complex multi-physical and multi-scale experimental and theoretical tools. Ageing tests may require long times and adequate acceleration strategies need to be developed; this could be a quite hard task since the proper accelerating parameters are not easy to single out. Moreover virtual testing through model simulations asks for comprehensive models, validated under a great variety of ageing conditions. The concern of the international community about the long term behavior of PMCs at high temperatures has led to the development of specific research programs and collaborations, bringing together industrial and academic partners. Recently, consistent research concerning the long term behavior of high glass transition temperature PMCs was carried out within the framework of the USA NASA High-Speed (HSR) and the French MENRT “Supersonique” research programs - both launched in the 1990s and aiming at the development of a 60000h-90000h long range supersonic aircraft. The present review chapter resumes the contribution of the authors to such research effort: • •
presenting some activities carried out within the context of the French MENRT “Supersonique” research program (2001-2004), collecting the results of a specific research action – the COMEDI research program (2005-2008) – consecrated to the study of chemo-mechanics couplings occurring during thermo-oxidation of PMCs at high temperatures.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
4
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
As mentioned, thermo-oxidation of polymer materials and PMCs at high temperatures represents a peculiar form of chemical (thermal) ageing: it is generally agreed that thermooxidation phenomena become relevant for temperatures higher than 120°C. In essence, thermo-oxidation intervenes at the macromolecular scale inducing chain scission, volatile departure, mass loss (thus material chemical shrinkage) and altering the mechanical properties of the oxidized material. In accord to several studies on unidirectional PMCs (see, for instance, [2-4]), it is now understood that substantial thermo-oxidation phenomena take place in material zones in the proximity of the external environment (some hundred of micrometers far from the external edges), while no significant degradation is found far from the external surfaces. Other research studies put into evidence the anisotropic thermo-oxidation behaviour of PMCs, showing faster oxygen diffusion along the fibres direction [5]. Classical modelling and simulation of thermo-oxidation relies on empirical models [6] but some important developments of the simplest models have been carried out [7-8] especially for PMR-15 resins and resin composites [9-10]; some models address the issue of damage and thermo-oxidation/damage interaction [11]. Some research carried out on PMCs laminates put into evidence the importance of thermo-oxidation on the onset and the development of micro (at the fibre level) and meso (at the ply level) damage on thermally cycled samples [12-14]. In recent years a thermo-oxidation mechanistic scheme for polymer matrix materials has appeared in the literature (see, for instance, [15-18]); this scheme is based on a kinetic model – a radical chain reaction – and is represented by a set of differential equations. Thermooxidation induced mass loss and polymer matrix chemical shrinkage have been also set on the basis of the predictions of the mechanistic scheme ([18]). By looking at the literature – with only few exceptions - two separate tendencies can be singled out: from one side, chemists have tried to build deterministic models to simulate the chemical reactions taking place during thermo-oxidation, from another side, researchers in mechanics have mainly focused their attention on the effects of thermo-oxidation on the mechanical properties of PMCs, while not much effort has been put forward trying to create a bridge between the chemical and the mechanical aspects of thermo-oxidation in PMCs. This link cannot be easily characterized since – as mentioned – all the involved fields (including the chemical and the mechanical ones) are coupled and it is not a simple task to write down such couplings without incurring in excessive generalizations and overwhelming difficulties. It is understood that a necessary starting point for dealing with thermo-oxidation in PMCs consists in studying the behaviour of the polymer resin alone and then extend the research to the PMC. Actually a tentative and comprehensive research program concerning thermo-oxidation in PMCs should be characterized – in its essential tracts - by the following steps: •
study and full characterization of the thermo-oxidation behaviour of the polymer resin, and identification of the main parameters affecting chemo-mechanics couplings in such a material. This is the simplest step, since this behaviour is expected to be isotropic. It should be noted that pure polymer resin samples are essential to characterize thermo-oxidation induced mass loss and matrix chemical shrinkage strain,
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures … •
•
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
•
5
study and full characterization of the thermo-oxidation behaviour of the PMC at the microscopic scale, that is, at the scale of its elementary constituents (fibres and matrix). In this study, the interphase zones and the fibre/matrix interfaces should deserve particular attention, since they could give rise to a complex behaviour. Fibres are scarcely and poorly reactive to thermo-oxidation phenomena: however – since fibres constitute a physical constraint to the free chemical shrinkage matrix strain consistent stress may arise at the fibre/matrix interface leading ultimately to fibre/matrix debonding. This is an essential (and unavoidable) feature of the thermooxidation behaviour of PMCs at the microscopic scale, study and full characterization of the thermo-oxidation behaviour of the PMC at the mesoscopic scale, that is, at the scale of the elementary PMC ply (the unidirectional lamina). In this case, the PMC material - viewed as a complicate aggregate of many fibres and the polymer – could be the place for complex anisotropic diffusion controlled oxidation. This behaviour can be characterized by means of homogenization procedures over a representative volume element (RVE); however since the oxidized zone extend over a few hundred of micrometers with consistent chemical and mechanical gradients - the notion of RVE is difficult to apprehend or even impossible to define. Since fibres constrain the free thermo-oxidation induced shrinkage of the matrix, the unidirectional lamina becomes the place for multiple strain (and) stress concentrations close to the fibre/matrix interfaces. The lamina itself should be affected by the thermo-oxidation induced chemical shrinkage of the matrix and should exhibit – at least locally – a free shrinkage strain directly proportional to that of the matrix, depending on the fibre volume fraction, study and full characterization of the thermo-oxidation behaviour of the PMC at the macroscopic scale, that is, at the scale of the laminate (sequence of unidirectional laminae). Mass loss and degradation can be appreciated at this scale. However, the eventual presence of diffuse damage may give rise to very complex interactive phenomena between chemical, mechanical and damage fields, leading to an acceleration of the mass loss kinetics. The mismatch of hygrothermal and thermooxidation induced properties may induce residual stress within a lamina.
It is clear that phenomena pertaining to a given scale may translate to another scale. For instance, damage onset phenomena at the microscopic scale may lead to damage onset and propagation phenomena at the meso or macro scale (through coalescence of microcracks, for instance). Therefore, the possibility of scale interaction and coupling phenomena should be taken into account. The present chapter – starting from the existing literature – presents a review of experimental and modeling strategies to deal with chemo – mechanics couplings in PMCs subjected to thermo-oxidation phenomena, at all the mentioned scales, and in particular at the microscopic scale. As mentioned, most of such developments were carried out within the context of the French COMEDI research program and some of them made the object of publications in congress and journal papers [19-22]. The review chapter is organised as follows: in section 1 thermo-oxidation phenomena are introduced; through a comprehensive review of the recent literature, some experimental facts
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
6
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
are discussed, giving also some details about the kinetic mechanistic model developed by Colin and Verdu [18]. In section 2 confocal interferometric spectroscopy methods are introduced as a tool to measure local thermo-oxidation induced shrinkage strains and deformations at the exposed edges of composite samples. In section 3 a multi-physics unified model approach based on the thermodynamics of irreversible processes is presented. The model is able to put in evidence chemo-mechanics couplings and shows how the oxygen reaction-diffusion within the polymer may be influenced by the mechanical variables. The classical mechanistic model for oxygen reactiondiffusion by Colin and Verdu [18] can be also recovered by introducing simplifying hypotheses. The model takes into account some “indirect” chemo-mechanics coupling - the elastic properties of the resin material are function of the local oxygen concentration - and the viscoelastic behaviour of the polymer at high temperature. In section 4 the model is identified and validated through different experimental tests and –once validated – satisfactorily employed for the simulation of thermo-oxidation induced local strains and stresses in composites under several different environmental conditions. Such simulations help identifying the critical conditions for thermo-oxidation induced damage onset and propagation. The possibility to accelerate thermo-oxidation ageing phenomena through increasing oxygen pressure is investigated and discussed in section 5 both experimentally and theoretically. Section 6 finally presents conclusions and perspectives. As a closing introductory remark - it has to be pointed out that research concerning thermo-oxidation in PMCs is far from being completed. Most of the research work performed by the authors of the present review chapter is still ongoing and will be the object of forthcoming communications and journal publications. From a general viewpoint, many achievements have been reached in this field of research so far, but much still needs to be done.
LITERATURE REVIEW: RELEVANT ISSUES AND EXPERIMENTAL FACTS The oxidation reaction occurring at the contact between the material and an oxidizing environment is a natural process in nature. In some materials, such as metals, oxidation can give rise to a protective layer (passivation) on the surface, which insulates the core material and prevents oxidation from degradation. This process typically occurs at room temperature and remains stable over time. In other cases, particularly for epoxy resins, oxygen penetrates into the material and diffuses towards its core - driven by concentration gradients - leading in some cases to the development of consistent oxidized zones, whose depth evolves with time. Though at room temperature the kinetics is slow and the entire process can eventually be neglected (unlike metal materials), thermo-oxidation takes significant proportions for temperatures close to around one hundred degrees. At such temperatures, an oxidized layer forms and rapidly progresses; within such layer the polymer (or PMC) mechanical properties
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
7
degrade and degradation may eventually lead to widespread damage in the form of matrix cracking, fibre/matrix debonding ... This phenomenon has been largely observed and documented in the literature; in the following subsections we will try to give some account of these observations, giving also some details about the classical mechanistic scheme developed by Colin and Verdu [18] to give a comprehensive interpretation of such phenomena. The literature review will begin with a short account about the effects of thermooxidation on damage onset and propagation in composite laminates: someone may argue that this is very complex issue to start with: however, the review of such issue will give some important information about the relevance of thermo-oxidation phenomena in PMCs materials and structures.
Effect of Thermo-Oxidation on Damage Onset and Propagation in Composite Laminates
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The role of thermo-oxidation on damage onset in [02/902]s composite laminates has been clearly demonstrated by performing thermal cycling tests between -50°C and 180°C under neutral (nitrogen) and oxidizing (atmospheric air) environments [1, 13]. Figure 1 shows SEM images of the external side edges of the off-axis (90°) plies of the laminate (after 1000 cycles). The investigation is carried out at the microscopic scale; the images illustrate a quite scarce number of fibres, separated by few resin rich pockets. The illustrations call for an important (though often underrated) remark; the fibre volume fraction of PMCs is often far from being uniform, even along quite small material zones. It is hard to find zones containing a regular/periodic fibre distribution; more often the resin rich areas are randomly distributed and are of very different size.
Figure 1. SEM images of the external side edges of the off-axis (90°) plies of [02/902]s composite laminates after 1000 thermal cycles between -50°C and 180°C under neutral (nitrogen, a) and oxidizing (atmospheric air, b) environments ([1, 13]).
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
8
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
The side surface of the specimen remains globally healthy in the case of thermal cycling in neutral environment (Figure 1a); on the contrary, consistent matrix shrinkage in low fibre volume fraction zones and multiple debonding along the fibre/matrix interfaces are visible on the side surfaces of specimens thermally cycled under oxidizing environment (Figure 1b). As mentioned, matrix shrinkage is in fact the result of successive chemical reactions leading to the generation of volatile species which - after leaving the polymer - contribute to local mass and density changes, therefore to an irreversible shrinkage volumetric matrix strain [18]. The onset of debonding is clearly the result of high local stresses at the fibre/matrix interfaces; these stresses are related to the thermal expansion mismatch between matrix and fibres, to the temperature differential (from 180°C to -50°C), to the thermo-oxidation induced matrix shrinkage strain and to the thermo-oxidation induced material property changes (for instance, antiplasticization effects, embrittlement of the interface [23]). The sample picks up thermo-oxidation effects during the high-temperature stage of the thermal cycle, at 180°C. Both samples are subjected to the same thermal effects and feel the same temperature differential; therefore, for the sample in Figure 1b, the accumulation of matrix shrinkage strain and the thermo-oxidation induced material property changes during the hightemperature stage of the cycle are really crucial for the onset and development of damage. It is evident from figure 1 that damage microcrack lips are open and constitute new surfaces for oxygen ingress and propagation into the material. Figure 2 shows X-ray images of [02/902]s composite laminates subjected to thermal cycles between -50°C and 180°C under neutral (nitrogen) and oxidizing (atmospheric air) environments [13]. This time the investigation is carried out at the meso/macroscopic scale; the images illustrate the whole composite specimen. In Figure 2, samples are disposed in such a way to show the external plies (0° plies) horizontally and the internal plies (90° plies) vertically. It can be seen that matrix mesocracks develop in both the external and the internal plies: cracks propagate from the external edges to the centre of the sample; moreover, crack density increases with increasing cycles.
Figure 2. Evolution of mesocracks in [02/902]s composite laminates subjected to thermal cycles between -50°C and 180°C under neutral (nitrogen) and oxidizing (atmospheric air) environments ([13]) .
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
9
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
However, while very few cracks develop under neutral environment, the samples cycled under oxidizing environment exhibit a much more damaged state and reach almost crack density saturation. These mesocracks are mainly related to the thermal expansion mismatch between adjacent plies, to the temperature differential (from 180°C to -50°C) and to the thermooxidation induced material property changes. Again, both samples are subjected to the same thermal history; the thermo-oxidation effects picked up at the high-temperature stage of the thermal cycle are crucial for the onset and the development of mesoscopic damage. In summary, figures 1 and 2 show the comparative effects of two distinct environments; in the first one (nitrogen) the whole thermal history and the quite dramatic temperature differential between high (180°C) and low (-50°C) temperatures are not sufficient to engender damage onset and propagation (at least at a significant extent), both at the microscopic and the meso/macroscopic scale; in the second one (atmospheric air), thermo-oxidation effects picked up during the high temperature phases of the cycle are clearly responsible for damage onset and propagation. Figures 1 and 2 demonstrate that chemo – mechanics – damage couplings should be taken into account for the understanding of damage evolution phenomena in PMCs. It is not much clear whether a relationship does exist between the damage state at the microscale and the one at the meso/macroscale: so far, it has been suggested that mesocracks could result from diffuse damage coalescence of microcracks (matrix flaws, fibre/matrix debondings); however, they could develop independently from microcracks, driven by the thermo-oxidation embrittlement of the matrix. Therefore the two phenomena could possibly result from two distinct (though similar) phenomena, acting at two distinct scales. Though this point has not been completely solved, it is clear that an investigation at the microscopic scale is of paramount importance to understand degradation phenomena in PMCs under thermo-oxidative environment. The present chapter will focus mainly on the effects of thermo-oxidation at the microscopic scale.
Effects of Thermo-Oxidation on the Neat Polymer The effects of thermo-oxidation on the behaviour of the neat resin material can be appreciated at two distinct scales: at the local (microscopic) scale through Ultra-MicroIndentation (UMI) measurements, at the global (macroscopic) scale through Dynamic Mechanical Analysis (DMA) measurements. UMI measurements are usually performed on small polymer resin samples (typical dimensions: 15 mm × 10 mm × 1 mm) and consist on enforcing a VICKERS-like indenter on the polished surface of such samples: the net surface of the indentation marks is around 2 µm2 and the distance between two adjacent measurement marks is typically 10÷20 µm. Through UMI tests one is typically able to measure – at room temperature - the VICKERS hardness (HV) and the Elastic Indentation Modulus (EIT), which is a measure of the local “elasticity” of the polymer material (for more details see reference [20]).
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
10
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
DMA measurements are usually carried out on standard polymer samples (average dimensions: 35 mm × 10 mm × 1 mm) and allow measuring – as a function of temperature – the elastic and the damping properties, namely the conservation (E’) and the loss moduli (E’’) of the polymer samples (for more details see references [21] and [43]). Figure 3 shows room temperature EIT profiles measured by UMI on 977-2 polymer resin samples subjected to an oxidizing environment (atmospheric air) at 150°C for three different durations - 100h, 600h and 1000h, respectively - as a function of the distance from the edge exposed to the environment. Figure 3 shows the same profiles measured on a sample conditioned under a non-oxidizing environment (vacuum) at 150°c for 1000h, serving as a reference, for comparison. These profiles clearly show that thermo-oxidation is driven by diffusion, since they reflect the fact that oxidation progresses from the exposed edge towards the centre on the sample. The affected zone is around 200 µm thick (from the exposed edge) and seems not varying with conditioning time. Furthermore, an EIT increase is systematically observed in the oxidized zone; this can be explained by invoking antiplasticization of the polymer material, a phenomenon widely detailed in [23], generated by the macromolecular chain scissions occurring during the thermo-oxidation chemical reactions. It should be noted that – for the test case and the material in figure 3 – the maximum EIT increase with respect to the virgin condition can be as high as around 30% after 1000h under atmospheric air environment, at 150°C. Similar profiles have been obtained for PMR-15 polymer resin materials oxidized at 315°C under atmospheric air [24].
Figure 3. Room temperature EIT profiles measured by UMI on polymer resin samples subjected to an oxidizing environment (atmospheric air) at 150°C for three different durations - 100h, 600h and 1000h. On the same curve EIT profiles measured on a sample conditioned under a non-oxidizing environment (vacuum) at 150°c for 1000h, serving as a reference, for comparison [20] .
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
Fiber straight
11
still
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 4. Schematics of typical DMA measurements (E’ and E’’) performed on thermo-oxidized (dotted lines) and non-oxidized (continuous lines) 977-2 polymer resin samples.
Figure 4 shows schematically the results of typical DMA measurements (E’ and E’’) performed on thermo-oxidized (dotted lines) and non-oxidized (continuous lines) 977-2 polymer resin samples (for more details see reference [21] and [43]). The thermo-oxidation induced antiplasticization effect [23] is represented by a horizontal shift and a diminution of the β-relaxation peak (occurring at around -60°C for such material) which leads to a vertical shift (increase) of the conservation modulus (E’) at room temperature. This result is coherent with the results of the EIT measurements. Another classical thermo-oxidation induced effect is represented by a horizontal shift of the polymer glass transition temperature, Tg, which can be appreciated on both the E’ and the E’’ curves. As mentioned, DMA measurements are conducted on polymer resin samples on a global scale. They give the response of the specimen and truly depend on the sample geometry and on the imposed mechanical solicitations. For instance – for samples subjected to three-point bending DMA tests – it is customary to measure a double peak for the E’’ curves in correspondence with the glass transition zone. This is typically due to the fact that it is almost impossible to oxidize a polymer resin sample up to saturation. The tested structure is thus characterized by two oxidized layers close to the external edges exposed to the environment and by an almost virgin zone at the samples heart; each zone is characterized by its own glass transition and the global sample response is equivalent to that of a “sandwich” material. These considerations put into evidence one of the main difficulties to be faced when dealing with the experimental characterization of the thermo-oxidation induced properties of polymer and PMCs materials. Since saturation is hardly attained, the oxidized polymer and PMCs material properties are not homogeneous along the samples and are thus difficult to identify through tests performed on samples. In the literature many experimental results are available concerning the mass loss of polymer resin samples; for instance, Decelle et al. [25] showed that the mass loss (∆m/m0) of 70 µm thick samples made of a mixture of aromatic epoxy crosslinked by an aromatic diamine can be as high as (around) 5% after 1000h exposition at 150°C under atmospheric air environment, leading to shrinkage strain values close to 2.5%.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
12
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Classical Mechanistic Scheme for Thermo-Oxidation in Polymers and PMCS Colin and Verdu [18] have developed a mechanistic kinetic scheme for modeling thermooxidation in polymers and PMCs, in order to reduce the level of empiricism of some existing models [6] and to describe in a detailed way the phenomena occurring during the thermooxidation processes. The mechanistic scheme is represented by a “closing loop” reaction producing its own initiator and can be schematically represented by the following set of reactions: POOH + γ PH → 2P° + H2O + νV (k1, initiation) P° + O2 → PO2° (k2, propagation) PO2° + PH → POOH + P° (k3, propagation) P° + P° → inactive products1 (k4, ending) P° + PO2° → inactive products2 (k5, ending)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
PO2° + PO2° → inactive products3 (k6, ending)
(1)
In such a scheme P represents the macromolecular chain, the symbol ° characterizes the free radicals and V represents an average molecule of volatile products. The process is based on the dissociation of POOH hydro peroxides in P° radicals, then forming PO2° radicals in the presence of oxygen; the radicals associate then with hydrogen atoms giving rise again to hydro peroxides. The scheme takes into account also some volatile species (H2O and V) and three distinct species of inactive products. The parameters k1 to k6 represent the reaction rates and are taken constant. Finally, γ and ν are adjustable parameters which have – however – a physical sense (for more details see [18]). By ignoring the diffusion of species other than oxygen and by assuming that the volatile products escape immediately from the material, the following system of nonlinear differential equations can be draft:
dC = - k2 [P°]C + k6 [PO2°]2 + DC ∇2C dt d[POOH] = k3 [PH][PO2°] – k1 [POOH] dt d[PH] = - k3 [PH][PO2°] – γ k1 [POOH] dt d[ PO 2 °] = k2 [P°]C - k3 [PH][PO2°] – k5 [PO2°][P°] – 2k6 [PO2°]2 dt
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
13
d[P] = 2k1[POOH] - k2 [P°]C + k3 [PH][PO2°] - 2k4 [P°]2 – k5 [P°][PO2°] dt
(2)
in which C is the oxygen concentration and DC is the oxygen diffusivity; the material is supposed to be isotropic with respect to the diffusion process. Moreover, in equation (2) the oxygen diffusion process is supposed to follow the Fick’s law. The system of equations (2) can be solved (numerically) by specifying the initial and boundary conditions; the oxygen equilibrium concentration at the exposed edges, Cs, can be related to the oxygen partial pressure of the ageing atmosphere, p, through the classical Henry’s law: Cs = pS
(3)
in which S is the coefficient of solubility of oxygen in the polymer. The local rate of oxygen consumption: r (C) = -
dC = k2 [P°]C - k6 [PO2°]2 dt
(4)
can be integrated over time to give the local amount of absorbed oxygen, Q (C): Q (C) =
∫
t 0
r (C) dτ
(5)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
It is worth noting that Q (C) is space and time dependent. Mass changes can be finally determined from a balance between weight gain due to oxygen consumption and weight loss due to departure of volatiles [18, 25]:
dC d[V] d[H 2 O] 1 dm 1 = (MC - MH2O - MV ) dt dt dt m 0 dt ρ0
(6)
in which MC, MH2O and Mv are, respectively, the O2, H2O and volatile products molar mass, m0 and ρ0 are the polymer initial mass and density, respectively. The shrinkage, Esh, is then given by: Esh =
1 ∆V 1 ∆m ∆ρ = ( ) 3 V0 3 m0 ρ0
(7)
in which V0 is the initial polymer volume. The reader is referred to [18, 25] for more details about equations (6) and (7). In the literature [18, 25], the reaction rate constants were identified (for several polymer resin systems) from mass loss curves measured on polymer resin samples with several thickness values and at different temperatures; the model was then validated on mass loss
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
14
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
curves measured on samples with thickness values different from those employed for identification purposes. A qualitative description of the solution of the system of differential equations (2) has been given by Colin and Verdu [15, 18], who also illustrated – for unidirectional diffusion, along the sample thickness (x – coordinate) - the qualitative shape of the function Q (x, t). First they noted that the local oxygen reaction rate depends nonlinearly on the local oxygen concentration, reaching an asymptotic value for high values of the latter. As a result of this phenomenon, in some cases, the Q (x, t) profiles may exhibit a shape qualitatively similar to that illustrated in figure 5 (for a sample with thickness e), in which the tangent to the curve close to the external exposed edges (x = 0 or x = e) is almost horizontal.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 5. Qualitative shape of the oxidation profiles Q (x, t) along the thickness direction (x coordinate) of a polymer resin sample with thickness e.
Figure 6. Numerical calculation of the local amount of absorbed oxygen, Q, as a function of the distance from the exposed edge for different conditioning times (a, atmospheric air, 150°C); room temperature EIT profiles measured by UMI on 977-2 polymer resin samples under the same environmental conditions (b) [20].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
15
As mentioned, the mechanistic scheme has been mainly validated by comparison with mass loss curves of polymer resin samples, at the global scale. However there exists an alternative validation method which consists in comparing the prediction of the mechanistic scheme with EIT profiles (such as those in figure 3), at the local scale [19, 20]. Figure 6a shows the numerical calculation of the local amount of absorbed oxygen, Q, as a function of the distance from the exposed edge for different conditioning times; simulations are run for a 1 mm polymer resin sample exposed to atmospheric air at 150°C. Figure 6b reproduces exactly figure 3, illustrating room temperature EIT profiles measured by UMI on 977-2 polymer resin samples under the same environmental conditions and for the same sample. The two figures are disposed side-by-side in order to illustrate the similar qualitative shape of the two profiles. It can be seen that the two curves follow closely the same temporal evolution and identify the same thermo-oxidized layer zone (around 200 µm): it can be deduced that the two properties are correlated. Figure 7 [20] illustrates the correlation that actually exists between the EIT and Q for each spatial point and at each time, at room temperature. Though phenomenological, this correlation is physically linked to the phenomenon of antiplasticization which has been proven to occur in polymer resin material systems [23]. Figure 7 is important for at least two reasons: • •
it shows that the mechanistic scheme is effective predicting thermo-oxidation in polymer materials (validation at the local scale), it allows identifying a phenomenological relationship between the EIT and Q.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The following functional relationship between and EIT and Q (at room temperature) can be identified for a 977-2 polymer resin: EIT (Q) = 5510 – 1469 exp (-0.48 Q) (MPa)
(8)
Figure 7. Correlation between the room temperature elastic indentation modulus (EIT) and the local amount of absorbed oxygen, Q [20].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
16
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Equation (8) represents a form of indirect coupling between diffusion and mechanics and summarizes the effect of oxygen reaction-diffusion on the mechanical properties of the polymer matrix. Despite its relevance at the local scale, the EIT (Q) cannot be actually seen as a pure measure of the Young modulus of a material. In fact EIT measures are strongly influenced by the test condition, in particular by the applied local force. However, the functional form EIT (Q) gives the relative local modulus variations with respect to a virgin condition, as a function of the local amount of absorbed oxygen. In order to find the room temperature engineering elastic constants of the polymer matrix material we start from the hypothesis – put forward by Verdu in [23] - that the bulk modulus of the polymer, K, is not affected by thermo-oxidation, that is, by Q. Then by measuring the room temperature Young modulus, E, and Poisson’s ratio, ν, of the virgin polymer material, an by hypothesizing that they follow both the same functional relation (8), the following formulas can be written, at room temperature [19]: E (Q) = 4422 – 1179 exp (-0.48 Q) (MPa) ν (Q) =
4422 - 1179 exp (-0.48Q) 1 (MPa) 2 19662
(9) (10)
from which the polymer shear modulus, G, can be evaluated – at room temperature - starting from the classical relation for isotropic materials.
Discussion on the Reviewed Experimental Facts
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
From the review of the literature, it is evident that a large body of work exists concerning thermo-oxidation in polymers and PMCs and many important results have been established over the years. In particular, it is today understood that: •
•
•
thermo-oxidation has many important effects on damage onset and propagation in composite materials and structures. It plays a fundamental role on the onset of fibre/matrix debonding - at the microscopic scale - and on acceleration of mesocracks growth and density increase – at the meso/macroscopic scale [1, 13]. thermo-oxidation engenders mass loss and matrix chemical shrinkage strain development. For instance, it has been shown [25] that the mass loss of polymer resin samples can be as high as 5% after 1000h exposition at 150°C under atmospheric air environment, leading to shrinkage strain values close to 2.5% [25], thermo-oxidation affects consistently the mechanical properties of the polymer material, its local “rigidity”, its glass transition temperature due to the effects of antiplasticization [23]. This can be measured at the local scale by UMI measurements and at the global scale by DMA measurements. For instance it has been noted that – locally – the EIT values of an oxidized resin may increase by around 30% with respect to the virgin material after 1000h under atmospheric air environment, at 150°C [20],
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures … •
17
a mechanistic kinetic scheme for thermo-oxidation of polymers and PMCs has been developed [18]. The model is able to reduce the level of empiricism of other existing models [6] and to describe in a detailed way the phenomena occurring during the thermo-oxidation processes. Moreover the mechanistic model introduces an important parameter, Q, which represents the local amount of absorbed oxygen. The parameters of the model are identified through comparison with mass loss curves [18, 25]. Then the model is validated against alternative mass loss curves (at the global scale) and by comparison with UMI-measured EIT profiles (at the local scale). Most importantly, through the employment of the model, a physically sounded phenomenological relationship between the EIT and Q can be established – at the local scale [20].
It is clear that the literature presents much consistent information about thermo-oxidation in polymers and PMCs. There exists still – however – a consistent lack of knowledge and in particular: •
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
•
•
•
•
no detailed quantitative studies exist concerning the development of thermooxidation induced irreversible matrix shrinkage deformations and strains in PMCs, at the local scale, such as those qualitatively observed by SEM (figure 1b). These deformation/strain fields are clearly responsible for damage onset at the microscopic scale (fibre/matrix debonding), thus their quantification (over time and all over the ageing process) is of paramount importance, there is no systematic study – both theoretical and experimental - concerning strong chemo-mechanics couplings in polymer and PMCs materials subjected to thermooxidative environments. Composite materials can be the place for consistent “internal” stresses – engendered by a mismatch of the basic constituents properties; it is important to demonstrate whether such stresses (besides those due to external applied forces) can be responsible for accelerating the kinetics of the oxygen reaction-diffusion process, there is no systematic study – both theoretical and experimental – concerning chemomechanics-damage coupling/interaction in polymer and PMCs. Though some research addresses this fundamental issue (see for instance [11, 26]) there is still no clear information about the onset of damage at the microscopic scale (figure 1b), the effect of thermo-oxidation on the development and acceleration of matrix mesocrack density at the mesoscopic scale (figure 2) and – most of all – no clear links between the two phenomena, there is also scarce information about the long term behaviour of real PMC structures. Recently, Cinquin and Medda [27] have presented some important results concerning mass loss changes and mechanical properties evolutions (open hole compression resistance) of long-term aged PMCs structures. These results still need an exhaustive interpretation, the fundamental issue of accelerating thermo-oxidation phenomena by means of temperature or pressure changes (increase) has still not found clear response.
The following sections of the present review chapter will try to give partial answer to these outstanding needs, in particular:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
18
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot •
•
• •
a novel experimental technique based on confocal interferometric microscopy (CIM) technique will be setup for the quantitative characterization of matrix shrinkage in PMCs, at the local scale, the bases for a fully coupled chemo-mechanics model for thermo-oxidation of polymers and PMCs will be presented. The model will be then validated against experimental observations - CIM measurements on composite samples and UMI in neat polymer resin samples subjected to the combined effect of strain gradient and thermo-oxidation. Some discussion on how the model could be employed for damage onset and propagation predictions will be also provided, the model predictions will be also compared to mass loss curves of PMCs structures, such as those in [27], experimental and theoretical tools for approaching and developing acceleration test techniques will be presented and discussed.
The results presented in the following sections made partly the object of a PhD dissertation [19] and have been presented in national and international colloquia [28-30]; journal papers from the research activity have been submitted for publication [31, 34]. As mentioned, the research is far from being complete. However, it is hoped that the partial results presented in this chapter will bring some light on thermo-oxidation phenomena in polymers and PMCs and will constitute a solid starting point for further speculations.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
CHARACTERIZATION OF MATRIX SHRINKAGE IN PMCS BY CONFOCAL INTERFEROMETRIC MICROSCOPY The thermo-oxidation induced matrix shrinkage profiles in PMCs samples (such as those qualitatively illustrated by SEM observations, figure 1b) can be quantitatively measured through confocal interferometric microscopy (CIM). In the present chapter these measurements have been done by employing a Taylor Hobson TALYSURF CCI 6000 microscope (see also [29, 30]). The method is based on the Michelson interferometry whose basic physical principles are briefly recalled: a ray of white light is split in two rays by a lens: the first ray is reflected by a reference mirror - placed at a fixed reference distance from the lens - the second ray is reflected by the sample - at a distance that can be changed by a piezoelectric actuator. The interference between the two signals – as captured by a CCD camera - is proportional to the difference between the two distances, so that - pixel by pixel - a chart of the vertical displacements of the sample can be quantitatively measured and the profiles obtained. Such an apparatus represents a non contact non destructing testing technique with a high vertical resolution (up to 0.01 nm), lateral resolution up to 0.4 µm. Due to technical limitations of the setup, the points (pixels) at which the slope of the measured displacements profiles exceeds the limiting value of 27° are reported as ‘unmeasured’ points.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
19
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 8. Example of CIM surface measurement on a PMC side surface.
Figure 8 gives some examples of the measuring technique and illustrates the displacement profiles of a thermo–oxidized PMC surface - measured on the polished sides quasi-isotropic carbon fibre reinforced polymer (CFRP) composite samples. Figure 8 shows over a surface of the order of a hundred micrometer square – the general features of surface measurement performed by the CIM technique: fibre rich zones (from which the local volume fraction can be measured), matrix rich zones and unmeasured points are clearly identified; quantitative displacement measurements are reported by the viewer as coloured contours. Over a composite sample edge several distinct zones – exhibiting quite different surface contours - can be singled out (figure 9): • • •
intraply matrix rich zones – resin rich pockets within a ply, characterised by a maximum fibre-to-fibre distance less than 50 µm, interply matrix rich zones – resin rich pockets at the interface between two distinct plies, characterised by fibre-to-fibre distances much higher than 50µm, intraply homogeneous zones, in which the fibre volume fraction is quite high and the matrix shrinkage effect is restrained by the dense concentration of fibres.
Intraply and interplies matrix rich zones exhibit a considerable amount of shrinkage; the observations are done at room temperature therefore shrinkage is due both to the oxidationinduced irreversible chemical strains and the reversible thermal strains picked up during the cool down from test temperature to room temperature. In both zones, matrix shrinkage profiles can be extracted along fibre-to-fibre paths; these profiles generally have the shape illustrated in figure 9 and give access to the maximum shrinkage depth along a path.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
20
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 9. Different types of surface which can be observed by CIM at the exposed edges of PMC samples: example of fibre-to-fibre profile and maximum shrinkage depth measurements.
Thermo-oxidation induced shrinkage profiles have been measured on the 90° plies of [+452/-454/+452/010/9010]s thick IM7/977-2 quasi isotropic (QI) PMCs samples. Samples were polished (with a polishing precision up to 1 µm) at the edges (where CMI observations were then made) and kept under oxidative environment (atmospheric air) at 150°C for 192h. Such an environment should be sufficiently aggressive to promote measurable amount of polymer matrix shrinkage but – at the same time – quite “soft” to prevent from fibre/matrix debonding and matrix cracking. As a general remark it is noted that the maximum shrinkage depth increases as the distance between fibres (fibre-to-fibre spacing) increases, that is, as the size of the matrix rich zone increases. The maximum measured shrinkage depth is around 2 µm for a fibre-to-fibre spacing equal to around 100 µm. Figure 10 shows the maximum thermo-oxidation induced shrinkage depth as a function of the fibre-to-fibre distance in matrix rich intraply and interply zones for unoxidized samples and samples aged 192h in atmospheric air at 150°C. It should be noted that unoxidized samples exhibit a certain amount of shrinkage – less than 0.5 µm - in a range of fibre-to-fibre spacing less than 50µm: this shrinkage is partly due to the reversible thermal strains picked up during the cool down to cure to room temperature and to the mechanical induced polishing strains.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
21
Figure 10. Maximum shrinkage depth as a function of the fibre-to-fibre distance for non-oxidized and thermo-oxidized PMCs samples (atmospheric air, 150°C, 192h).
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
In the thermo-oxidized PMC sample matrix shrinkage is due to the superposition of mechanical induced polishing strains, reversible thermal strains picked up during the cool down to cure to room temperature and irreversible chemical strains due to thermo-oxidation at 150° (192h). The dotted line situated at a fibre-to-fibre distance equal to 50 µm separates the measurements performed on intraply matrix rich zones by those performed on interply matrix rich zones, that is, along ply-to-ply interfaces.
CHEMO-MECHANICS COUPLED MODEL FOR THERMO-OXIDATION OF POLYMERS AND PMCS A chemo-mechanics coupled model of thermo-oxidation is developed in order to: • • • • •
give satisfactory interpretation of the experimental CIM observations presented in the preceding section for PMCs samples, establish a rationale to predict the onset of damage at the microscale (fibre/matrix debonding), promote experiments to catch diffusion-mechanics couplings in polymers and PMCs, establish prediction and simulation methods for mass loss in PMCs structures, promote experiments for accelerated thermo-oxidation in polymers and PMCs.
The model is based on the Thermodynamics of Irreversible Processes (TIP) developed by De Donder and Prigogine [35-37]. A similar formalism has been employed for the modeling
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
22
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
of coupled heat transfer, mechanics and gas diffusion in polymers in [38], for the modeling of coupled mechanics and water diffusion in epoxy matrices [39] and for the modeling of reaction-diffusion-heat-mechanics couplings during the cure of epoxy matrices [40]. The model takes advantage of the experimental observations presented in the preceding sections. The model is actually split in two parts: first we consider strong chemo-mechanics couplings in materials in the elastic range, then we model – separately – the viscoelastic behaviour of the polymer at high temperature. In the following subsections we sketch the fundamental features of the model, preceded by a short account of the thermodynamics framework in which the theory is set; full details can be found in references [31-40].
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Remarks on the Thermodynamics Framework The Thermodynamics of Irreversible Processes (TIP) is founded on the fundamental contributions by Duhem, Onsager, De Donder and Prigogine ([35]) and deals with the dissipative macroscopic behaviour of solids out-of-equilibrium. It is a phenomenological approach in which, however, strict and rigorous connections with the behaviour of matter at the microscopic scale (statistical mechanics) have been provided over the years. In TIP is fundamental the notion of state, represented by a collection of a certain number of variables which may be observable (and eventually controllable) or internal (hidden). At equilibrium, a thermodynamic state can be clearly identified by the clear definition (and full meaning) of all the state variables taken into account: for instance – at equilibrium – the notion of temperature makes perfect sense. Moreover – at equilibrium – a certain number of thermodynamic potentials can be defined and the Gibbs equation is satisfied. Rigorously – within a body in equilibrium – the spatial distribution of the state variables should be uniform and homogeneous; for instance, a body at equilibrium should have constant and homogeneous temperature. When a collection of states is considered (thermodynamic process) the notion of equilibrium is lost for two reasons: the body is in a transient condition (the state variables are changing with time) and may loose its homogeneity (the state variables may change from point to point giving rise to spatial gradients). TIP is based on the following postulates: •
•
the state variables are defined as local quantities; they exist at each point of the body and have full meaning; for instance, a local absolute temperature and specific entropy can be defined at a material point of a body, independently of the other points; at each instant of time, each material point is seen as if it was in thermal equilibrium; in particular, the thermodynamic potentials can be defined and the Gibbs equation is satisfied, locally.
This constitutes the basis of the axiom of local state.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
23
Therefore - in TIP - the equations of state can be determined by a thermodynamic potential (the specific free energy, ψ, for instance) defined locally over the entire set of the state variables. Moreover – following TIP – a “rate of specific entropy”, ds/dt, can be defined for a material point - given by the sum of an exchange term, des/dt and of an internal term, dis/dt, that is:
ds d s d s = e + i dt dt dt
(11)
The internal term is due to internal dissipation due to irreversible phenomena, the exchange term includes heat flux terms that compensate the entropy variation promoted by internal dissipation. According to the second principle of thermodynamics, the internal dissipation term must be greater than zero for irreversible transformation, while it is equal to zero for reversible transformations. According to TIP, the dissipation, Φ, is given by: Φ=ρT
d is = y · z& ≥ 0 dt
(12)
in which ρ is the density, T the absolute temperature, y a vector of generalized thermodynamic forces and z& a vector of generalized velocities (or “fluxes”). Since TIP is concerned with the evolution of material systems close to equilibrium - it is almost natural to assume that generalised forces and velocities are related by a linear affine homogeneous relationship (Onsager-Casimir reciprocity relation):
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
z& = L · y
(13)
in which L is a non singular tensor whose “symmetry” has found experimental support in many branches of physics [35]. Equation (12) allows writing:
z& =
∂ D* (y) ∂y
(14)
where D* (y) is the Legendre-Frenchel transform of a dissipation potential D ( z& ) and is a quadratic non-negative form (pseudo-potential). By this assumption a thermodynamic process becomes admissible, since Φ is consequently non-negative. In conclusion – in TIP – the material behaviour is governed by the two non-negative functions, the thermodynamic potential, for instance ψ, and the dissipation pseudo-potential D*.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
24
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Chemo-Mechanics Couplings in the Elastic Range Full development of the present section can be found in reference [32]. Consistently with TIP, the fully coupled chemo-mechanics model is built by defining proper dissipation and thermodynamic potentials. In doing so, an important step consists identifying the proper state variables and fluxes. We identify first the proper balance equations for the polymer material. We suppose that an elementary volume of polymer matrix behaves as a continuum and is represented by a perfect homogeneous mixture of polymer and mobile chemical species. The mass balance of each ith mobile specie, of mass fraction Yi, within an elementary volume can be written as (see, for instance, [35]):
ρ
∂Yi = ∂t
nr
∑
νir Mi wr - ∇ · j mi
(15)
r =1
in which ρ is the density, νir the stoechiometric coefficient of the rth reaction, Mi the molar mass of the ith specie, wr the rth reaction rate, j mi the mass flux of the ith species and nr the total number of reactions. Equation (15) may be written in an equivalent form:
∂Yi* = ∂t
nr
∑
νir wr -
r =1
1 ∇ · j mi Mi
(16)
*
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
by introducing the molar concentration Yi = ρYi/Mi, i.e. the number of moles per unit volume. Indicating with X the position of a material particle in the reference configuration, with x its position in the actual configuration, the displacement u can be then defined as: u=x–X
(17)
and, within the context of a small strain theory, the strain E is given by: E=
1 ∇⊗u + ∇⊗uT) (∇ 2
(18)
For small strains, the actual and the reference configurations can be confused and the Cauchy stress tensor S is subjected to the balance equation: ∇·S+f=0
(19)
in which f are generalized body forces. The stress and strain tensors, S and E respectively, are decomposed into their spherical (Ss, Es) and deviatoric (Sd, Ed) components:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures … Ss =
1 trS I; Sd = S – Ss 3
Es =
1 trE I; Ed = E – Es 3
25
(20)
The strain/stress decomposition is motivated by the hypothesis – formulated by Verdu [23] at the molecular scale and mentioned in the preceding sections - that the bulk modulus K of the polymer matrix is not influenced by thermo-oxidation; therefore, the mechanical response of the material can be split in two contributions, a first one governed by the spherical component of the stress/strain tensor - unaffected by thermo-oxidation - the other governed deviatoric component of the stress/strain tensor – affected by thermo-oxidation through the matrix shear modulus G = G(Q). Moreover we note that in a linearized setting trE = trEan + trEe + trET + trEH + trESH, that is, the trace of the total strain tensor (E) is equal to the sum of the traces of the elastic strain tensor (Ee), the anelastic strain tensor (Ean) and the thermal (ET), hygroscopic (EH) and irreversible chemical shrinkage free strain tensors (ESH), respectively. In turn the free strain tensors can be related to the respective volume relative variations (∆V/V0)β and to the respective Jacobian, Jβ, by relations of the type trEβ = (∆V/V0)β = Jβ – 1. Linear relations are usually employed so that trEβ =
∑
γβ ∆β, in which γβ are
i
coefficients of thermal, hygroscopic or chemical expansion and ∆β is the related variation (β = T, H or SH). It should be noted that in the case of thermo-oxidation induced shrinkage ESH can be taken equal to Esh I, in which Esh is the shrinkage given by the chemical mechanistic model, equation (7). Following these developments, the trace of the total strain tensor, trE, its deviatoric part, Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Ed, the mass fraction, Yi (or, equivalently, the molar concentration,
Yi* ) can be chosen as
state variables. The specific Helmoltz free energy per unit mass (J/kg), ψ, is then taken quadratic, convex with respect to the state variables and to the internal variables (if any) and concave with respect to the temperature; in the present case – since we deal with isothermal processes - the temperature dependency will not be written explicitly: *
ψ = ψ (trE, Ed, Yi )
(21)
In this way, by deriving ψ with respect to such state variables, the trace of the Cauchy stress tensor, trS, its deviatoric part, Sd, and the chemical potential, µi, can be obtained, respectively. The Cauchy stress can be seen as the “force” driving the deformation process; in turn, the *
chemical potential - the derivative of ψ with respect to Yi - can be seen as a sort of “force” driving the diffusion processes. The Gibbs equation can be written:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
26
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot ns
∑
1 1 d Tds = de trS dtrE S : dEdρ ρ
niv
∑
µi dYi +
i =1
fj · dvj
(22)
j=1
in which T is the absolute temperature, s the specific entropy per unit mass (J/kgK), e the specific internal energy per unit mass (J/kg), µ i is the chemical potential of the ith specie (J/kg). vj represents a set of internal variables, fj are the associated thermodynamic “forces”, ns is the total number of chemical species and niv is the total number of internal variables. In equation (22) the term
∑
µi dYi can be also expressed by the equivalent
∑
i
ρ-1 µi d Yi in *
*
i
which µi = µi Mi is the chemical potential of the ith specie per unit mole (J/mol). The second principle can be expressed as follows: *
ρ
ds 1 = (- ∇ · q + r dt T
ns
∑
nr
µi (
i =1
∑
νir Mi wr - ∇ · jmi ) + ρ
r =1
niv
∑
fj ·
j=1
dvj ) dt
(23)
in which q the heat flux and r the internal heat source. According to Prigogine et al. [35] the entropy variation can be decomposed into an exchange term and internal term, that is:
ρ
d es 1 =−∇·( (q− T dt
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
d s 1 1 +( ) ρ i = q· ∇ dt T T
ns
∑ i =1
ns
nr
∑
µi jmi )) +
Ar wr -
r =1
∑
jmi · ∇
i =1
1 r T
µi ρ + T T
niv
∑ j=1
fj ·
d vj dt
(24)
In equation (24) Ar is defined by: ns
Ar = -
∑
ns
νkr Mk µk = -
k =1
∑
νkr µk *
(25)
k =1
and represents the “affinity”, in the sense of De Donder ([35]). In the present context internal dissipation terms due internal variables can be discarded, therefore, by introducing the entropy flux term jS as follows:
jS =
1 (q− T
ns
∑
µi jmi )
i =1
the dissipation, Φ, is finally given by:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(26)
27
Effects of Thermo-Oxidation on Composite Materials and Structures … ns
nr
Φ = - jS · ∇ T +
∑
Ar wr -
r =1
∑
jmi · ∇ µI
(27)
i =1
Equation (27) shows that dissipation is related to spatial gradients of temperature or chemical potential and to the chemical reactions. From equation (27) the generalized thermodynamic forces (∇ ∇T, ∇ µi, Ar) and the generalized velocities (or “fluxes” jS, wr, jmi) can be easily recognized giving rise to the pdeudo-potentials of dissipation: D = D (jS, wr, jmi) D* = D* (- ∇T, Ar, − ∇ µi)
(28)
By defining an equivalent strain E* = Ed : Ed the following quadratic dissipation potential can be chosen:
D* =
+
1 2
1 1 ∇T) + (-∇ ∇T) · BT (trE, E*) · (-∇ 2 2
ns
∑
(-∇ ∇µi) · Bµ i (trE, E*) · (-∇ ∇µi) +
i=1
n s −1 ns
∑∑
nr
∑
Br (trE, E*) Ar2 +
r =1
(-∇ ∇µi) · Cµ ij (trE, E*) · (-∇ ∇µj) (29)
i =1 j=1
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
in which BT, Br, Bµ and Cµ ij are strain dependent coefficients associated, respectively, to heat transfer, chemical reaction and diffusion. The reaction rate and mass flux are then given by:
wr =
∂D* = Br (trE, E*) Ar = - Br (trE, E*) ∂Ar
jmi =
ns
∑
ns
νkr Mk µk = - Br (trE, E*)
k=1
∂D* ∇µj) = - Bµ i (trE, E*) · ∇µi - Cµ ij (trE, E*) · (-∇ ∇µi) ∂ (-∇
∑
νkr µk *
k=1
(30)
Equation (30b) can be also re-written as: jmi = -
* Bµ i (trE, E*) * Cµ ij (trE, E ) * · ∇ µi · ∇µj Mi Mj
(30c)
Equation (30a) expresses the reaction rate wr as a function of the chemical potential. Equation (30b-c) relates the mass flux to the gradient of the chemical potential. It has to be noted that by equations (29) and (30b-c) the mass flux of the ith species is related not only to the gradient of its own chemical potential (Fick’s first law) but also to the gradient of the chemical potential of the jth species (with j ≠ i). The tensor Cµ ij relates the mass flux of the ith species to the gradient of the chemical potential of the jth species.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
28
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot Substitution (30a) and (30b) into (16) leads to:
∂Yi* = ∂t
ns
nr
∑
νir (- Br (trE, E*)
r =1
∑
νkr µk ) *
k =1
*
*
Bµ i (trE, E ) Cµ ij (trE, E ) * * 1 1 ∇ · (· ∇ µi ) · ∇µj) ∇ · (M M i j Mi Mi
-
(31)
Equation (31) expresses the general form of the mass balance equation for each ith mobile specie, dependent on the chemical potential per unit mole, µi . Obviously, equation (31) must be solved with the adequate boundary conditions, which will be discussed later. *
The final form of equation (31) depends upon the choice of the chemical potential µi , *
which, in turn, depends on the thermodynamic potential, ψ. An appropriate choice for ψ can be the following:
3 * * 1 ( K (trE)2 + G(Q( Yi )) E*) + ψ (trE, Ed, Yi ) = 2 ρ ns
∑ i =1
1 ( Ci trE ρ
Yi* + αi (trE, E*) RT ( Yi* (ln ( Yi* / Yi*0 ) – 1)) + µi0* Yi* )
(32)
in which K is the bulk modulus, G(Q(Y*i)) the reaction dependent shear modulus, αi (trE, E*), Ci are chemo-mechanics coupling coefficients to be identified. In particular αi (trE, E*) is a strain dependent coefficient related to the solubility, which play a role in the sorption process, as we will see later. Finally µi is a reference chemical potential which may depend on temperature and pressure (standard conditions) but does not depend on molar concentration
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
0*
*0
*
*0
and Yi is a reference molar concentration so that the ratio ( Yi / Yi ) corresponds to the chemical activity of the ith species. It should be noted that, in equation (32), the molar concentration
Yi* corresponds to the chemical activity of the ith species by assuming a *0
*
reference molar concentration Yi equal to 1 (mol m-3 if Yi is expressed in mol m-3). The state laws, including an expression for the chemical potential, can be recovered from ψ, as follows:
trS = Sd =
∂ρψ = 3K trE + ∂ trE
ns
∑ i =1
∂αi (trE, E*) * * ( RT (Y*i (ln ( Yi ) – 1)) + Ci Yi ) ∂ trE
∂ρψ * = 2G (Q( Yi )) Ed + 2 ∂ Ed
ns
∑ i =1
(
∂αi (trE, E*) * * RT ( Yi (ln ( Yi ) – 1))) Ed ∂ E*
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
29
∂ G (Q(Y*i)) * * 0* µ*i = ∂ρψ* = E + Ci trE + αi (trE, E*) RT ln Yi + µi ∂Y ∂ Y*
(33)
i
i
By substituting equation (33c) into equation (25) the affinity Ar becomes: ns
Ar = -
∑
νkr µk = *
k =1
ns
∑
νkr (
k =1
∂ G (Q(Y*k)) * * 0* E + Ck trE + αk (trE, E*) RT ln Yi + µk )(34) * ∂Yk
By substituting equation (33c) into (31) the mass balance of each ith mobile specie, of *
molar concentration Yi , within an elementary volume can be written as: ns ∂Yi* nr ∂ G (Q(Y*k)) * * 0* * = νir (- Br (trE, E ) νkr( E + Ck trE + αk (trE, E*) RT ln Yk + µk ))+ * ∂ Y k ∂t r=1 k =1
∑
∑
Bµ iαi (trE, E ) ∂ G (Q(Y i)) * 1 ∇ · (·∇( E + Ci trE + αi (trE, E*) RT ln Mi ∂ Y* i Mi *
-
*
Cµ ijαi (trE, E ) ∂ G (Q(Y j)) * * 0* 1 ∇ · (·∇( E + Cj trE + αj (trE, E*) RT ln Yj + µj ))(35) Mj ∂ Y*j Mi *
-
Yi* + µi0*)) +
*
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Equation (35) describes general reaction-diffusion of the ith species and is characterized by the sum of three terms. The first one is the reaction term; the second one is a diffusion term dependent on the gradient of the chemical potential of the ith species itself, the third one is a diffusion term depending on the gradient of the chemical potential of the jth species. In particular, concerning the reaction part, •
Br (trE, E*) is a reaction coefficient which may depend on the strain tensor,
•
the term
ns
∑ k=1
νkr
∂ G (Q(Y*k)) * E follows from the dependency of G on Q thus on ∂ Y*k
Yi* ,
•
which has been proven experimentally in the preceding section. This term must be present at least theoretically, the term Ck trE is related to a chemical shrinkage strain term in equation (33a) whose existence is proved experimentally; therefore this term should be present at least theoretically,
•
the term αi (trE, E*) RT ln Yi is specific to the chemical reaction and is essential to
*
construct the mechanistic scheme, as it will be shown later. It should be noted that αi may depend on strain. The diffusion part is characterised by the spatial gradients of both chemical and mechanical quantities, still involving the deviatoric part and the trace of the strain tensor, E*
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
30
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
and trE respectively, and a classical diffusion term, depending on the species concentration,
Yi* . The last diffusion term, involving the strain dependent tensor Cµ ij, is relevant when coupling between the fluxes of different mobile chemical species does exist. Boundary conditions for equation (35) can be found by imposing - at the interface between the environment and the “solvent” material - the equality of the chemical potentials of the gaseous species and of the species dissolved within the material. The chemical potential of the gaseous species, µ*g, can be classically written:
µ*g = µg0* + RT ln (p/p0)
(36)
0* where µg is the gas reference potential, p is the gas pressure and p0 is a reference pressure.
The chemical potential µis of a species dissolved within the material at the interface with the *
*
environment, ( Yis ) can be written:
µ*is =
∂ G (Q(Y*is )) * * *0 0* E + Cis trE + αis (trE, E*) RT ln ( Yis / Yis ) + µis ∂ Y*is
(37)
0* We remark that when equation (37) do not depend on E* and trE – and with µg = µis –
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
0*
we may recover interface boundary conditions which are analogous to the classical Henry’s law. As a last general remark we note that E* and trE may be the effect of an external applied strain/stress or the result of the self-generated strains/stresses related to the reaction-diffusion process itself through the free chemical shrinkage strain, ESH. The former case is usually referred in the literature as stress assisted diffusion. The second case is usually referred in the literature as self assisted diffusion. In order to recover the mechanistic scheme by Colin and Verdu [18] - equation (2) - and a classical boundary condition of the Henry type – equation (3) - some assumption should be made, that is: • •
only one mobile species must be considered, thus the tensor Cµ ij must be zero. the chemical reaction is not affected by mechanics, therefore the Br and the Bµ i coefficients are constant
•
the terms
•
to be negligible, the coefficients αi (trE, E*) do not depend on the strain components.
ns
∑ k =1
νkr
∂ G (Q(Y*k)) * E and Ck trE and their spatial gradients are assumed ∂ Y*k
Full development of equations (35) and (37) in the uncoupled case – and related discussion – can be found in reference [32].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
31
Effects of Thermo-Oxidation on Composite Materials and Structures …
Starting from equation (35), the mechanistic scheme by Colin et al. [18] can be generalized taking into account coupling with mechanics. Without loosing generality, only oxygen diffusion is considered: therefore Cµ ij =0 and Bµ O2 ≠ 0. By indicating with [O2] the oxygen concentration , equation (28) becomes: n
s ∂ G (Q(Y k)) * * 0* ∂[O 2 ] ={νO2-2 (- B2 (trE, E*) νkr( E + Ck trE + αk(trE, E*) RT ln Yk + µk )) + * ∂Y k ∂t k =1
*
∑
ns
+ νO2-6 (- B6 (trE, E*)
∑
νkr(
k =1
∂ G (Q(Y*k)) * * 0* E + Ck trE + αk (trE, E*) RT ln Yk + µk ))}+ ∂ Y*k
Bµ O2 (trE, E ) ∂ G (Q([O2])) * 0* + ∇·∇( E + C O2 trE + αO2 (trE, E*) RT ln [O2] + µ O2 )) (38) MO22 ∂ [O2] *
Again equation (38) must be solved by imposing opportune boundary conditions. In equation (31) chemo-mechanics couplings appear in several different forms. The dependency of G on Q([O2]), which is proven experimentally makes the term ∂ G (Q([O2])) * E appearing in equation (38). ∂ [O2] However, since E* follows from a product of strain tensors and we are within the framework of the small strain hypothesis, these terms can be neglected in a first approximation. Moreover the variation of G with respect to the concentration of oxidation products (equation (8)) is quite weak. ∂ G (Q([O2])) * By ignoring the E term equation (38) and its boundary condition equation ∂ [O2] (37) can be written, respectively:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
∂[O 2 ] = {νO2-2 (- B2 (trE, E*) ∂t
ns
∑
νkr(Ck trE + αk (trE, E*) RT ln Y*k + µ0*k)) +
k =1
ns
*
+νO2-6 (- B6 (trE, E )
∑
νkr(Ck trE + αk (trE, E*) RT ln Y*k + µ0*k))}
k =1
*
+
Bµ O2 (trE, E ) ∇ · ∇ (C O2 trE + αO2 (trE, E*) RT ln [O2] + µ0* O2)) MO22
µ*[O2]s = C[O2]s trE + α[O2]s (trE, E*) RT ln ([O2]s /[O2]0s) + µ0*[O2]s
(39) (40)
Chemistry kinetics is dependent on trE and E* through the coefficients B2, B6 and αi. On the other hand diffusion kinetics is influenced by mechanics through the term Bµ O2. The trE term influences directly the chemical reaction through the Ck trE terms and modifies the diffusion path through the ∇ · ∇ (C O2 trE) term. The trE term influences also the boundary condition through the C[O2]s coefficient. Equations (39) and (40) present a certain degree of complexity and must be solved numerically also due to the presence of a boundary condition explicitly depending on strain.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
32
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
At this stage of the theoretical development we may think about possible experimental tests to be done in order to quantify thermo-oxidation chemo-mechanics couplings. Two “families” of tests should be conceived; the first test should be done by imposing a uniform strain field on the material sample in order to evaluate the effect of the magnitude of the trE and E* terms on the oxygen reaction-diffusion. This test should be, in turn, performed on thin and thick samples in order to check the effect of mechanics on the sorption process (sample saturation) and on the diffusion process, respectively. A second test should be performed on samples subjected to strain gradients in order to enhance the effect of such gradients on the diffusion path. As shown in the literature review section, EIT measurements are able to catch experimentally the evolution of the Q spatial profiles; they have been proven effective for validating the classical thermo-oxidation mechanistic scheme. The idea is to employ the same technique in samples thermo-oxidized under an applied external strain/stress. Results from an experimental activity on chemo-mechanics couplings employing EIT measurements in polymer thermo-oxidized polymer resins will be presented in the next subsection.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Experimental Assessment of Chemo-Mechanics Couplings in Neat Polymer Resins under Stress In this sub-section the predictions of the chemo-mechanics fully coupled model will be tested against experiments. In fact, the model predicts a possible dependence of the oxygen reaction-diffusion process on the trace of the strain tensor, trE, and its spatial gradients (see equations (39) and (40), for instance). Full details about this sub-section can be found in reference [33]. The relevance of such predictions is studied experimentally through comparison with dedicated tests. In particular, two different tests are performed: a first test involves the relative magnitude of the trE term; another test involves the spatial gradients of trE. Homogeneous unnotched and notched neat polymer samples under homogeneous tensile longitudinal strain and exposed to a thermo – oxidizing environment are employed, in order to enhance - in the first case - the effect of the trace of the strain tensor and – in the second case - the effect of its spatial gradients. The effects of the reaction-diffusion process are then assessed at room temperature by ultra – micro – indentation EIT profile measurements, following the experimental procedure evocated in the preceding sections. Before chemo-mechanics coupled testing, resin samples were fabricated according to an optimized curing cycle characterized by a gelation phase (3h, 150°C, 7b), a vitrification phase (2h, 180°C, 7b) and a post-curing phase under vacuum (1,5h, 210°C). In this way the material reached a stable condition and all changes occurring during ageing were related to the exposition to the environment, almost exclusively. In a first test homogeneous unnotched 977-2 neat resin polymer samples were exposed to a thermo – oxidizing environment (48 hours at 150°C under 5b O2 - corresponding approximately to 1000h under atmospheric air at the same temperature) and subjected simultaneously to a uniform – displacement controlled - longitudinal strain equal to 2% and 4%, that is, at around 30% and 60% of the failure strain, respectively.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
33
Figure 11 shows Indentation Elastic Modulus profiles at room temperature as a function of the distance from the exposed surface.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 11. Room temperature EIT profiles measured by UMI in unnotched 1 mm thick 977-2 resin samples subjected to tensile strain (0%, 2%, 4%) (at 150°C under 5b O2, 48h) .
It can be noted that - taking into account the experimental dispersion of results - the applied uniform strain has no effect on the measured EIT profiles; it can be concluded that a homogeneous strain does not perturb significantly the diffusive mass flux and any change induced by such strain is of the second order. In order to put in evidence the effects of the stress gradients, notched 977-2 resin samples were fabricated. The schematic geometry of such samples is reported in figure 12. This sample has a section reduction equal to 1/5 in correspondence with the notch; therefore – in this zone – the resulting strain is highly heterogeneous even for an homogeneous applied strain at the boundaries. The sample was exposed to a thermo – oxidizing environment, 48 hours at 150°C under 5b O2 and subjected simultaneously to a uniform – displacement controlled – longitudinal strain. Actually, the sample was first charged up to around 84N and - once this value of force attained - the displacement control was imposed and kept constant. For illustration, the trace of the strain tensor field was numerically calculated (by employing the FE commercial code ABAQUS® [42]) for a value of the external applied force equal to 84N and by assuming a purely elastic behaviour for the resin: figure 13 reports graphically the results of such calculation.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
34
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Figure 12. Geometry of the notched 1 mm thick 977-2 resin sample.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 13. Illustration of the trE field calculated (ABAQUS® [42]) for a notched 1 mm thick 977-2 resin sample subjected to a tensile solicitation (84N).
Figure 14. Room temperature EIT profiles measured by UMI in notched 1 mm thick 977-2 resin samples subjected to tensile strain (at 150°C under 5b O2, 48h). Measurements are taken at points 1, 2, 3, 4 in figure 13.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
35
The strain field is highly heterogeneous and several zones along the sample can be identified for testing the relevance of the diffusion-mechanics coupling: in zones close to the notch (points from 1 to 3 in figure 13), the gradient is quite consistent and clearly related to the section reduction; in a zone sufficiently far from the notch (point 4 in figure 13), the field becomes homogeneous and the gradients become very weak. In order to assess the relevance of couplings, EIT profiles have been measured at points from 1 to 4 going from the exposed surface to the centre of the sample, following the direction indicated in figure 6 by black arrows. The results of such measurements (an average of at least three tests) are reported in figure 14, as a function of the distance from the exposed surface, along the different paths. The profiles are qualitatively similar to those obtained for the unnotched sample subjected to a homogeneous strain field. Then – and most importantly - the profile measured at point 4 is not significantly different from those measured at points from 1 to 3, within a zone which is affected by strong strain gradients. By such test, it is clearly shown that the effects of chemo-mechanics couplings are of the second order. Starting from these experimental results, the strong (direct) chemo-mechanics coupling terms in equations (39) and (40) can be neglected, at least in a first approximation and for the material under consideration.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Viscoelastic Model of a Polymer at High Temperature Since direct (strong) chemo-mechanics coupling can be neglected in a first approximation, the mechanics dependent term in equations (39) and (40) can be discarded and the oxygen reaction-diffusion model reduces to the classical mechanistic scheme by Colin and Verdu [18], equation (2) with boundary condition (3). Therefore, this section focuses on the viscoelastic behaviour of the polymer resin material at high temperature, neglecting strong chemo-mechanics couplings. Full details of the developments of the present section can be found in reference [31]. As a general introductory remark for the development of a viscoelastic model within the contex of TIP it is noted that deviations from equilibrium (which are small in TIP) and irreversible processes can be characterized, for instance, by a set of internal variables, whose instantaneous values during the non equilibrium evolution of the system are sufficient to define the state of the system itself, as if it was in local equilibrium. Internal variables are almost fictitious variables introduced to describe physical phenomena which usually take place at a microscopic scale and that would be intricate to model in all their complexity. In the present context the mechanical equilibrium state of the system is seen as a “relaxed” state, which is supposed to exist. The system is brought out of equilibrium via external perturbation, external stimuli such as forces or internal stresses, etc; then it comes back to equilibrium via a series of elementary mechanisms described by j generalized internal variables zj, to which a characteristic time τj is associated. These out of equilibrium processes (that tend towards equilibrium) are seen as linear, in the spirit of classical TIP. Such an approach, formally introduced by Cunat [41] in a largest contest, has been employed for the modelling of water diffusion – mechanics couplings (see, for instance, reference [39]).
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
36
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Two kinds of internal variables are introduced: scalar internal variables associated to the spherical part of the strain tensor, ztrEj, and tensorial internal variables associated to the deviatoric part of the strain tensor, ZEdj. The specific Helmoltz free energy per unit mass (J/kg), ψ, is here chosen as thermodynamic potential and taken quadratic, convex with respect to the state variables (the trace of the strain tensor, trE, and its deviatoric part, Ed) and to the internal variables (ztrEj, ZEdj) and concave with respect to the temperature; the temperature dependency is again discarded (isothermal conditions). The thermodynamic potential ψ has finally the form: ψ = ψ (trE, Ed, ztrEj, ZEdj)
(41)
and, explicitly: ρψ =
3 3 K∞ (trE)2 + G∞ (Ed : Ed) 2 2
n
∑
m
∑
Kj (ztrEj - z∞trEj)2 -
j=1
Gj (ZEdj - Z∞Edj)2 (42)
j=1
where K∞, G∞, z∞trEj and Z∞Edj represent relaxed values of the bulk modulus, the shear modulus and of the internal variables, respectively. The dual variables associated to state and internal variables can be found by derivation: ∂ρψ ∂ρψ ∂ z∞ + = 3K∞ trE ∂ trE ∂ z∞trEj ∂ trE trEj
trS =
∂ρψ ∂ρψ ∂ Z∞ S = + ∂ Ed ∂ Z∞Edj ∂ E
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
∑
3Kj β trEj (ztrEj - z∞trEj)
j=1
m
Edj
d
n
d
= 2G∞E -
∑
2Gj β Edj (ZEdj - Z∞Edj)
j=1
AtrEj = -
∂ρψ = 3 Kj (ztrEj - z∞trEj) ∂ ztrEj
A Edj = -
∂ρψ = 2 Gj (ZEdj - Z∞Edj) ∂ ztrEj
(43)
where AtrEj and AEdj can be seen as thermodynamic “forces” (scalar and tensorial, respectively) associated to the respective internal variable. The general expression of the Gibbs fundamental equation is: Tds = de -
1 1 trS dtrE - Sd : dEd + ρ ρ
∑
AtrEj d ztrEj +
j
∑ j
and the second principle of thermodynamics takes the form:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
AEdj : d ZEdj
(44)
Effects of Thermo-Oxidation on Composite Materials and Structures … ds 1 = (- ∇ · q + r + ρ dt T
ρ
∑
AtrEj
j
d ztrEj +ρ dt
∑
AEdj :
j
d ZEdj ) dt
37
(45)
In the spirit of TIP, the entropy variation can be then decomposed into an exchange term, des, and internal term, dis, that is: ρ
ρ
des 1 =−∇·( (q− T dt
dis 1 ρ = q· ∇ + ( ) dt T T
∑
AtrEj
j
∑
1 r T
µi jmi )) +
i
d ztrEj ρ +( ) T dt
∑
AEdj :
j
d ZEdj dt
(46)
The dissipation Φ takes the form: Φ=ρΤ
dis 1 = T q· ∇ + ρ dt T
∑
AtrEj
j
d ztrEj +ρ dt
∑
AEdj :
j
d ZEdj dt
(47)
and, by introducing the entropy flux term jS: jS =
1 (q− T
∑
µi jmi )
(48)
i
finally: Φ = - jS · ∇ T + ρ
∑
AtrEj
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
j
d ztrEj +ρ dt
∑
AEdj :
j
d ZEdj dt
(49)
The dissipation function and its Legendre-Frenchel transform are then: D = D (jS, ż trEj, ŻEdj) D* = D* (-∇ ∇T, ρ A trEj, ρ AEdj)
(50)
and, more precisely: D* =
1 1 (-∇ ∇T) · BT · (-∇ ∇T) + 2 2
∑
BtrEj (ρ A trEj)2 +
j Edj
1 2
: (ρ AEdj)
∑
(ρ AEdj) :
j
(51)
in which BT, BtrEj and Edj are coefficients associated, respectively, to heat transfer and to the relaxation phenomena. The rates of the relaxation processes can be then expressed by:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
38
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot dz trEj ∂ D* = = BtrEj (ρ A trEj) dt ∂ (ρ A trEj) dZ Edj ∂ D* = = dt ∂ (ρ AEdj)
Edj
: (ρ AEdj)
(52)
that, combined with the state equation (43) give: dz trEj = 3Kj ρ BtrEj (ztrEj - z∞trEj) dt dZ Edj = 3Gj ρ Edj : (ZEdj - Z∞Edj) dt
(53)
or, equivalently: dz trEj = (τ trEj )-1 (ztrEj - z∞trEj) dt dZ Edj =( dt
Edj -1
) : (ZEdj - Z∞Edj)
(54)
The internal variables at equilibrium z∞trEj and Z∞Edj are defined in such a way that their variation (at equilibrium) is related to the variation of the state variables through a set of generalised rigidities, β trEj and β Edj: dz∞trEj = β trEj d(trE)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
dZ∞Edj = β Edj d(Ed)
(55)
The spherical and the deviatoric parts of the stress tensor become then: n
trS = 3 (K∞ +
∑
n
Kj (β trEj)2) trE - 3
j=1
∑
Kj β trEj ztrEj
j=1
n
Sd = 2 (G∞ +
∑ n
Gj (β Edj)2) Ed - 2
j=1
∑
Gj β Edj Z Edj
j=1
and, by defining the glass bulk and shear moduli of the polymer matrix material as: n
KV = (K∞ +
∑
Kj (β trEj)2)
j=1
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(56)
Effects of Thermo-Oxidation on Composite Materials and Structures …
39
n
GV = (G∞ +
∑
Gj (β Edj)2)
(57)
j=1
finally: n
trS = 3 KV trE - 3
∑
Kj β trEj ztrEj
j=1
n
S d = 2 GV E d - 2
∑
Gj β Edj Z Edj
(58)
j=1
The parameters to be identified are the glass bulk and shear moduli of the polymer matrix material, KV, GV, then, for each j, the parameters Kj, Gj, β trEj, β Edj and the relaxation operators (τ trEj )-1 and ( Edj )-1. The last operator is here taken isotropic.
Identification of the Viscoelastic Model for a 977-2 Polymer Matrix Resin at High Temperature
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The viscoelastic behaviour of the polymer matrix material at high temperature has been identified through uniaxial tensile tests at 150°C performed on 977-2 resin samples instrumented with strain gages. Three values of longitudinal strain, respectively 0.8%, 1.8% and 2.2% were imposed instantaneously on the samples and kept for 18 hours: during the test the longitudinal strains Exx, the transverse strains Eyy and the longitudinal force (thus stress Sxx) were measured giving access to the relaxation laws of the materials. For such a test the relevant relationships needed for identification purposes are: trS = Sxx trE = Exx + 2 Eyy Sxy =
Sxx 2
1 Exy = (Exx - Eyy) 2
(59)
n
Sxx = 3 KV (Exx + 2 Eyy) - 3
∑
Kj β trEj ztrEj
j=1
n
Sxy = 2 GV Exy- 2
∑
Gj β Edj Z Edj
j=1
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(60)
40
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
and: dz trEj = (τ trEj )-1 (ztrEj - z∞trEj) dt dz Edj = (τ Edj )-1 : (ZEdj - Z∞Edj) dt
(61)
Following Cunat [41] a further simplification is added to the model by writing: Kj = pjK KR Gj = pjG GR
(62)
Equation (62) relates the j bulk and shear moduli Kj, Gj to two unique parameters, the “relaxed” bulk and shear moduli KR, GR through the some weight parameters pjK and pjG for which:
∑
pj K = 1
j
∑
pj G = 1
(63)
j
The weight parameters can be related to the relaxation times through the relationship:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
pjK = ((τ trEj )0.5 + G
Edj 0.5
pj = ((τ
)
+
k trE )/ (τ trEj )0.5 k (τ
)
((τ trEp )0.5 +
k trE ) (τ trEp )0.5
((τ Edp )0.5 +
k Ed ) (τ Edj )0.5
p=1 n
Ed
Edj 0.5
n
∑
)/
∑ p=1
(64)
in which ktrE and kEd are parameters that modify the weight repartition on the different relaxation times, eventually enhancing the relative importance of the small ones. The distribution (64) has been chosen in order to obtain a good agreement with the experiments and not starting by physical considerations, as in Cunat [41]. The relaxation times spectrum is defined by a maximum value, τ max and extended over a number of decades N; in the present study, 50 characteristic relaxation times will be considered. The same distribution of relaxation times is taken for the bulk and deviatoric evolutions (NtrE = NEd and τ trEmax = τ Edmax = τ max). Moreover the coefficients β trEj and β Edj have been taken equal to 1. Table 1 presents list of the identified viscoelastic model parameters for a 977-2 polymer resin material at 150°C. Figures 15 shows, respectively, the experimental values of trS and Exy measured during the relaxation tests and the correlation with the numerical viscoelastic model.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
41
In particular, for the unoxidised resin matrix at 150°C, KV (150°C) = 1400 MPa and GV (150°C) = 850 MPa, from which EV (150°C) = 2121 MPa. Table 1. Viscoelastic parameters for a 977-2 polymer resin material at 150°C
Specific parameters Global parameters
Viscoelastic bulk behaviour KV = 2395 KR = 900 ktrE = 650 N=4 τ max = 105
Viscoelastic deviatoric behaviour GV = 850 GR = 390 kEd = 400
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
(a)
(b) Figure 15. Measured and calculated stress (trS and Sxy) evolution during relaxation tests for three different imposed longitudinal strain values (0.8%, 1.8%, 2.2%).
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
42
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
VALIDATION OF THE MODEL AND SIMULATIONS The present section provides validation of the thermo-oxidation model against experimental characterization. In particular, in the following sub-section, we present validation of the thermo-oxidation model through comparison with matrix shrinkage and mass loss measurements. The model will be then employed for stress simulation under the action of aggressive environment. Full detail about the developments presented in the present section can be found in references [31] and [34].
Validation of the Model through Comparison with Cim Matrix Shrinkage Measurements in PMCs
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
In this sub-section the predictions of the chemo-mechanics coupled model will be tested against CIM matrix shrinkage measurements in PMCs. The viscoelastic behaviour of the 977-2 polymer resin material at high temperature is taken into account by the model equations (41)-(58), identified at 150°C (see table 1). The model is solved numerically by employing the ABAQUS® finite element commercial code [42], through the employment of dedicated subroutines. The C3D20T element is employed for simulations. Full details about the numerical implementation of the model can be found in reference [19]. Figure 16 shows a schematic illustration of the geometrical configurations adopted for simulations. A square packed fibre – matrix representative geometry model (fig.16a) is employed for intraply matrix rich zones where full 3D effects are expected. A quasi-2D representative model geometry (Figure 16b) is employed for interply matrix rich zones (plyto-ply interfaces) in order to enhance the expected – almost plane strain – effects. The schematics in figure 16 show also the point at which the shrinkage is calculated.
Figure 16. Representative model geometries for numerical simulations.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
43
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 17 shows the boundary conditions employed for the oxygen reaction – diffusion and the mechanical problem, respectively. The oxygen concentration is kept constant on the surface in contact with the gaseous environment (top edge surface), in which no mechanical constraints are specified; on the other surfaces, the oxygen flux is set equal to zero and opportune displacements are set in order to respect the symmetry conditions of the representative cell. The representative model geometry is made sufficiently long (> 200 µm) in order to simulate correctly the reaction – diffusion problem and not to suffer from constraints coming from the bottom surface. The loading history of the transient simulations is as follows: at the start time, the sample is free from stress and strain at his curing temperature (210°C), the temperature is then lowered instantaneously (in one static step) to 150°C and kept to this value all along the oxidation phase. Finally – at the end time - the temperature is again instantaneously lowered to room temperature. This loading history aims at simulating the sample behaviour all along a real thermooxidation test. The temperature difference between cure temperature and room temperature simulates – though in an approximate manner – the contribution coming from residual curing stresses. Figure 18 illustrates a comparison between the experimental measured and the simulated maximum shrinkage depth as a function of the distance between adjacent fibres; the polymer matrix composite sample is aged under oxidative environment (atmospheric air) at 150°C for 192h. Simulations concern both intraply and interply zone. Simulations reproduce closely the experimental trends; the maximum shrinkage depth increases with increasing fibre-to-fibre distance.
Figure 17. Boundary conditions imposed on the model geometries.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
44
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Figure 18. CIM measured and numerically simulated maximum shrinkage depth as a function of the fibre-to-fibre distance in thermo-oxidized samples (atmospheric air, 192h, 150°C).
The behaviour of interply matrix rich zones is slightly different from that of interply matrix rich zones, since these two zones are subjected to a consistently different constraint conditions. This motivates and justifies the need for an opportune geometry model capable to simulate different geometric constraining effects.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Model Simulations – Micro Damage Onset Once the model is validated, it is employed in a predictive manner to simulate thermooxidation induced shrinkage strain and stress for different fibre-to-fibre spacing values and different environmental conditions, including the reference case of neutral environments. The thermo-oxidation induced shrinkage and stress are simulated in PMCs samples under aggressive thermo – oxidative (5b O2, 48 hours, 150°C) and neutral environments (5b N2, 48 hours, 150°C), respectively. The geometry employed for simulations is representative of a simplified square packed fibre – matrix arrangement in which the elementary cell is 4µm × 4µm and the fibre diameter is 6µm. Figure 19 shows the evolution of the shrinkage displacement at point P1 as a function of the simulation time and the environment. The instantaneous cooling from the stress free cure temperature (210°C) to the test temperature (150°C) is simulated in one static step and produces a displacement which is common to the two environments. Thermo-oxidation shrinkage strains and displacements develop during the transient phase (up to 48h) in samples under thermo – oxidative environments; the response of the two samples tend to diverge after a certain “induction” time, whose duration is influenced by the type and the temperature of the oxidative environment.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
45
Figure 19. Transient shrinkage displacement of point P1 for a PMC sample under thermo-oxidative and neutral environments.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
At room temperature, the sample under thermo-oxidative environment has picked up more shrinkage displacement (around 48% relative increase) than the sample under neutral environment. Figure 20 shows the evolution of the Von Mises stress at point P1 as a function of the simulation time and the environment. During the thermo-oxidation phase at 150°C, stress relaxation at high temperature takes place for both samples: however, while the sample under neutral environment keeps relaxing up to the end of the oxidation phase, the sample under thermo-oxidative environment starts develop stress after the “induction” time. This stress is related to the shrinkage strain and to the material properties changes occurring in the neat polymer during thermo-oxidation.
Figure 20. Transient Von Mises stress of point P1 for a PMC sample under thermo-oxidative and neutral environments.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
46
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 21. Adimensional Von Mises stress (average along path P1-P2) as a function of the fibre-to-fibre distance.
At room temperature, the sample under oxidative environment has picked up more Von Mises stress (around 150% relative increase) than the sample under neutral environment; in fact, - within the induction time - a competition takes place between stress relaxation due to viscoelasticity and stress building due to thermo-oxidation. Figure 21 shows the adimensional Von Mises stress (averaged along the path P1 – P2) as a function of the fibre-to-fibre distance for a PMC under thermo-oxidative environment (5b O2, 48 hours, 150°C). The average Von Mises stress increases with increasing fibre-to-fibre distance. According to such predictions damage onset should rather take place in configurations with high fibre-to-fibre spacing (low local volume fraction) than in zones with high local volume fraction. Figure 21 allows making only qualitative predictions about the onset and the eventual progression of thermo-oxidation induced damage in PMCs. Strain and stress predictions should be coupled to an opportune damage onset criterion in order to follow thermo-oxidation-damage diffusion and propagation in PMCs exposed to aggressive environments over long times. This should lead to the prediction of the durability performance of PMCs based structures. Work is in progress in order to elucidate this important topic.
Mass Loss Simulation in PMCs Laminates The ageing behaviour (mass loss) of two PMC samples (A and B) at 150° C under atmospheric air environment has been simulated by the numerical model implemented in ABAQUS®. Both samples are unidirectional: in sample A fibres are along the length direction while in sample B fibres are along the thickness direction; the geometry is the same for the two
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
47
samples (100 mm × 30mm × 4 mm) therefore the surface exposed to the oxidative environment is identical. The diffusion constant for the composite are respectively D11 = Dresin = 1.3 µm2/s in the fibre direction and D22 = D33 = 0.8 µm2/s in the direction transverse to the fibres. Only 5000 hours of ageing were simulated. The simulated mass loss values are compared to the experimental ones in Figure 22. The experimental trend is well reproduced by the model. The experimental and numerical simulated values have the same order of magnitude. The mass loss curves simulated for sample B are very close to the experimental ones up to 2000 hours and tend to diverge thereafter; this finding should be related to the observation of some interlaminar matrix cracks appearing on the oxidized surfaces of sample B (starting from 1000 hours) and whose number increases thereafter. The creation of new surfaces for oxygen ingress could explain the differences between experimental and simulated mass loss; in its present version, the model does not take into account the phenomenon of damage onset and growth in composite laminates.
ACCELERATED THERMO-OXIDATION
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Accelerated thermo-oxidation tests are needed in order to reduce the costs related to longterm sample exposure and ageing. However, accelerating thermo-oxidation is a quite complex issue, due to the complexity and multiplicity of mechanisms involved in thermo-oxidation phenomena.
Figure 22. Measured and predicted mass loss of PMCs samples subjected to thermo-oxidative environment (atmospheric air, 150°C).
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
48
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Increasing test temperature represents an interesting way to accelerate ageing; several different ageing phenomena in polymer resin materials follow an Arrhenius-like temperature dependence. However exposure at high temperature may promote unwanted degradation, in particular additional crosslinking and related chemical changes. In most polymer resins the temperature range between the application temperature (at which thermo-oxidation occurs) and the glass transition temperature (at which additional crosslinking starts taking place) may be very narrow, so that acceleration by means of temperature increase becomes unpractical. Another way to accelerate thermo-oxidation consists in increasing the partial pressure of the environment oxygen, by employing, for instance, pure oxygen instead of atmospheric air; by imposing a higher oxygen partial pressure on the sample external surface, a higher oxygen concentration will result on such surface, depending on the material solubility. Pressure induced accelerated thermo-oxidation will be reported in the present chapter – the result of several investigation studies (see, for instance, reference [34]). Industrial resin samples were aged at 150°C under 5 bar pressure under both neutral and oxidizing environments (for 18h, 48h, 96h and 430h). Figure 23 shows the EIT profiles as a function of the distance from the sample free edge, for different aging times and for the two different environments. No significant EIT changes (with respect to the virgin material) were measured in samples aged neutral environment (Figure 3a). On the contrary, a systematic increase (with respect to the virgin material) of the EIT profiles was measured in samples under oxidizing environment (Figure 23b), all along the thermo-oxidized layer. The thickness of such layer (around 200 µm) was found almost identical to that measured on samples aged under atmospheric air. It should be noted that - for samples oxidized under 5b O2 - the EIT profile flattens near the edge of the sample along a distance lower than 50 µm. This effect can be related to the nonlinear behaviour of the chemical reaction leading to saturation of the reaction rate under oxygen pressure increase.
Figure 23. Room temperature EIT profiles measured by UMI on polymer resin samples under (a) neutral (5b N2) and (b) oxidizing environment (5b O2) at 150°C for different durations. On the same curve the typical room temperature EIT values for the unoxidized resin, serving as a reference.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
49
The measured EIT and their spatial distribution were correlated to the concentration of grafted oxygen (Q) profiles calculated with the reaction-diffusion mechanistic model, equation (2). Instead of equation (3), the following boundary condition was employed for the oxygen concentration at the external surface, Cs: ' Cs = kd p + C H bp
1 + bp
(65)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
in which p is the environment gas pressure and kd is a solubility-like coefficient to be identified (in the present study kd = 0.03001 mol l-1 bar-1). CH’ and b are two other parameters related to a Langmuir-type behaviour to be identified (in the present study CH’ = 0.0003 mol l-1 and b = 0.848 bar-1). Equation (3) expresses the classical sorption Henry’s law, which is adapted for low partial pressures: on the contrary, equation (65) represents a dual-mode-like sorption law [44], which is more appropriate for relatively high pressures and, in particular, for the present application. Figure 24 shows the correlation between the room temperature measured EIT and the concentration of oxidation products (Q) curves, for both samples aged under atmospheric air and samples aged under 5b of pure oxygen. In the figures all tests are collected, namely, 100h, 600h and 1000h ageing under atmospheric air and 18h, 48h, 96h and 430h under 5b O2. All data belong to a unique low scattered curve having the following phenomenological form, which is identical to equation (8). This phenomenological law expresses a physical mechanism relating the EIT to the Q values within the material. The existence the same correlation law for different values of pressure is a major result and is essential for the understanding of pressure accelerated thermo-oxidation phenomena.
Figure 24. Correlation between room temperature EIT profiles measured by UMI and Q profiles in polymer resin samples aged under atmospheric air and 5b O2.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
50
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
CONCLUSION The present chapter has illustrated some research activity carried out by the members of the Physics and Mechanics Department – Insitut Pprime – ENSMA and concerning the effects of thermo-oxidation in composite materials and structures at high temperatures (T > 120°C). Thermo-oxidation phenomena have been first introduced within the context of a comprehensive literature review, illustrating the relevant issues and some experimental facts. Confocal interferometric spectroscopy methods have been then presented as a tool to measure local thermo-oxidation induced shrinkage strains and deformations at the exposed edges of composite samples. A multi-physics unified model approach based on the thermodynamics of irreversible processes has been then presented; this model is a generalization of a classical mechanistic model for oxygen reaction-diffusion and includes strong chemo-mechanics couplings and includes the possible influence of the mechanical variables on the oxygen reaction-diffusion phenomena. Starting from the predictions of the model, chemo-mechanics coupled tests have been conceived and carried out. According to experimental observation it was actually proved that the mechanical variables have no effect on the oxygen reaction-diffusion process, at least in a first approximation, for the solicitations and the materials under investigation. The model has been then identified and validated through comparison with polymer matrix shrinkage local measurements and PMCs sample mass loss: once validated has been satisfactorily employed for the simulation of thermo-oxidation induced local strains and stresses in composites under several different environmental conditions. The tendency for thermo-oxidation induced damage onset at the microscopic scale has been clearly illustrated by the model simulations, though the analysis still rests on almost qualitative grounds. The possibility to accelerate thermo-oxidation ageing phenomena through increasing oxygen pressure has been finally investigated and discussed both experimentally and theoretically. The investigation at the meso/macroscopic scale is still far to be completed. Future research activity will concern the study of thermo-oxidation induced damage onset and propagation at such scale.
ACKNOWLEDGMENT Part of the developments presented in this chapter has been carried out within the framework of the COMEDI research program, funded by the French ANR RNMP agency. All partners of the research, PIMM-ARTS ET METIERS ParisTech Paris and EADS IW Suresnes are gratefully acknowledged. Noel Brunetiere (LMS University of Poitiers) is also acknowledged for his help in performing the confocal interferometric microscopy measurements.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Effects of Thermo-Oxidation on Composite Materials and Structures …
51
REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
[14] [15] [16] [17] [18] [19] [20] [21]
[22] [23] [24] [25]
Lafarie-Frenot, MC. International Journal of Fatigue, 2006, 28, 1202-1216. Magendie, F; Seferis, J; Aksay, I. 35th International SAMPE Symposium, 1990, 22802288. Parvatareddy, H; Wang, JZ; Lesko, JJ; Dillard, DA; Reifsnider, KL. Journal of Composite Materials, 1996, 30, 210-230. Madhukar, M; Bowles, KJ; Papadopoulos, DS. Journal of Composite Materials, 1996, 31, 596-618. Scola, D. Proceedings of the Joint U.S.-Ital Symposium on Composite Materials. Plenum Press, NY, 1983, 159-169. Bowles, KJ; Nowak, G. Journal of Composite Materials, 1988, 22, 966-985. Tsotsis, TK. Journal of Composite Materials, 1998, 32, 1115-1135. Wang, SS; Chen, X. Journal of Engineering Materials and Technology, 2006, 128, 8189. Tandon, GP; Pochiraju, KV; Schoeppner, GA. Materials Science and Engineering: A 2008, 498, 150-161. Pochiraju, KV; Tandon, GP; Schoeppner, GA. Mechanics of Time Dependent Materials, 2008, 12, 45-68. Pochiraju, KV; Tandon, GP. Composites Part A: Applied Science and Manufacturing 2009, 40, 1931-1940. Lafarie-Frenot, MC; Rouquie, S. Composites Science and Technology, 2004, 64, 17251735. Rouquie, S; Lafarie-Frenot, MC; Cinquin, J; Colombaro, AM. Composites Science and Technology, 2005, 65, 403-409. Lafarie-Frenot, MC; Rouquie, S; Ho, NQ; Bellenger, V. Composites Part A: Applied Science and Manufacturing, 2006, 37, 662-671. Colin, X. Ph.D. Dissertation, 2000, ENSAM Paris, France. Colin, X; Verdu, J. Revue des Composites et des Matériaux Avancés, 2002, 12, 63-186. Colin, X; Marais, C; Verdu, J. Polymer Degradation and Stability, 2002, 78, 545-553. Colin, X; Verdu, J. Composites Science and Technology, 2005, 65, 411-419. Olivier, L. Ph.D. Dissertation 2008, ENSMA Poitiers, France. Olivier, L; Ho, NQ; Grandidier, JC; Lafarie-Frenot, MC. Polymer Degradation and Stability, 2008, 93, 489-497. Rasoldier, N; Colin, X; Verdu, J; Bocquet, M; Olivier, L; Chocinski-Arnault, L; Lafarie-Frenot, MC. Composites Part A: Applied Science and Manufacturing, 2008, 39, 1522-1529. Olivier, L; Baudet, C; Bertheau, D; Grandidier, JC; Lafarie-Frenot, MC. Composites Part A: Applied Science and Manufacturing, 2009, 40, 1008-1016. Pascault, JP; Sautereau, H; Verdu, J; Williams, RJ. J. Thermosetting polymers, 2002, Marcel Decker Inc; New-York. Johnson, LL; Eby, RK; Meador, MAB. Polymer, 2003, 44, 187-197. Decelle, J; Huet, N; Bellenger, V. Polymer Degradation and Stability, 2003, 81, 239248.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
52
Marco Gigliotti, Jean-Claude Grandidier and Marie Christine Lafarie-Frenot
[26] Colin, X; Mavel, A; Marais, C; Verdu, J. Journal of Composite Materials, 2005, 39, 1371-1389. [27] Cinquin, J; Medda, B. Composites Science and Technology, 2009, 69, 1432-1436. [28] Grandidier, JC; Olivier, L; Gigliotti, M; Lafarie-Frenot, MC; Vu, DQ. CFM’09 XIXème Congrès Français de Mécanique 2009, Marseille, France. [29] Gigliotti, M; Vu, DQ; Olivier, L; Grandidier, JC; Lafarie-Frenot, MC. JNC16 XVIèmes Journées Nationales sur les Composites, 2009, Toulouse, France. [30] Gigliotti, M; Vu, DQ; Olivier, L; Lafarie-Frenot, MC; Grandidier, JC. ICCM17 XVII International Conference on Composite Materials, 2009, Edinburgh, UK. [31] Gigliotti, M; Olivier, L; Vu, DQ; Grandidier, JC; Lafarie-Frenot, MC. Journal of the Mechanics and Physics of Solids, 2011, 59, 696-712. [32] Gigliotti, M; Grandidier, JC. Comptes Rendus de l’Académie des Sciences – Mécanique, 2010, 338, 164-175. [33] Gigliotti, M; Grandidier, JC; Lafarie-Frenot, MC. Mechanics of Materials, 2011, 43, 431-443. [34] Lafarie-Frenot, MC; Grandidier, JC; Gigliotti, M; Olivier, L; Colin, X; Verdu, J; Cinquin, J. Polymer Degradation and Stability, 2010, 95, 965-974. [35] Prigogine, I; Kondepudi, D. Thermodynamique 1999, Editions Odile Jacobs, Paris. [36] Lebon, G; Jou, D; Casas-Vasquez, J. Understanding Non-Equilibrium Thermodynamics 2008, Springer-Verlag, Berlin. [37] Germain, P; Nguyen, QS; Suquet, P. ASME J. Appl. Mech., 1983, 50, 1010-1020. [38] Rambert, G; Grandidier, JC. European Journal of Mechanics A/Solids, 2005, 24, 151168. [39] Valancon, C; Roy, A; Grandidier, JC. Oil & Gas Science and Technology–Rev. IFP 2006, 61, 759-764. [40] Rabearison, N; Jochum, C; Grandidier, JC. Computational Materials Science, 2009, 45, 715-724. [41] Cunat, C. Mechanics of Time Dependent Materials, 2001, 5, 39-65. [42] ABAQUS 6.7 Users Manual. [43] Chocinski-Arnault, L; Olivier, L; Lafarie-Frenot, MC. Materials Science and Engineering A 2009, 521-522, 287-290. [44] Klopffer, MH; Flaconneche, B; Oil and Gas Science and Technology–Rev IFP 2001, 56, 223-244.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
In: Composite Materials in Engineering Structures Editor: Jennifer M. Davis, pp. 53-136
ISBN: 978-1-61728-857-9 © 2011 Nova Science Publishers, Inc.
Chapter 2
DAMPING IN COMPOSITE MATERIALS AND STRUCTURES Jean-Marie Berthelot1, Mustapha Assarar2, Youssef Sefrani3 and Abderrahim El Mahi4 1
ISMANS, Institute for Advanced Materials and Mechanics, 44 Avenue Batholdi, 72000 Le Mans, France 2 University of Reims, GRESPI, 9 Rue de Québec, 10026 Troyes, France 3 Faculty of Mechanical Engineering, University of Aleppo, Syria 4 University of Le Maine, LAUM, Avenue O. Messiaen, 72085 Le Mans, France
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
1. Introduction Damping in composite materials is an important feature of the dynamic behaviour of structures, controlling the resonant and near-resonant vibrations and thus prolonging the service life of structures under fatigue loading and impact. Composite materials generally have a higher damping capacity than metals. At the constituent level, the energy dissipation in fibre-reinforced composites is induced by different processes such as the viscoelastic behaviour of matrix, the damping at the fibre-matrix interface, the damping due to damage, etc. At the laminate level, damping is depending on the constituent layer properties as well as layer orientations, interlaminar effects, stacking sequence, etc. The initial works on the damping analysis of fibre composite materials were reviewed extensively in review papers by Gibson and Plunket [1] and by Gibson and Wilson [2]. Viscoelastic materials combine the capacity of an elastic type material to store energy with the capacity to dissipate energy. The most general treatment has been given initially by Gross [3] considering the various forms that viscoelastic stress-strain relations can take. A form of the viscoelastic stress-strain relations is that involving the complex moduli, where the stress field is related to the strain field introducing a complex stiffness matrix. Thus, the static elastic solutions can be converted to steady state harmonic viscoelastic solutions simply by replacing elastic moduli by the corresponding complex viscoelastic moduli. The elasticviscoelastic correspondence principle was developped by Hashin [4, 5] in the case of
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
54
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
composite materials. Furthermore, Sun et al [6] and Crane and Gillespie [7] applied the correspondence principle to the laminate relations derived from the classical laminate theory. Following this process, the effective bending modulus of a laminate beam can be derived [8], leading to the experimental evaluation of laminate damping. This complex modulus was also considered by Yim [9]. Indeed, the experimental analysis implemented in the case of unidirectional glass and Kevlar fibre composites [10] shows that the complex stiffness model leads to a rather worse description of the experimental results derived for the damping as a function of fibre orientation. A damping evaluation of composite materials has been developed initially by Adams and Bacon [8] in which the energy dissipation can be described as separable energy dissipations associated to the individual stress components. This analysis was refined in later paper of Ni and Adams [11]. The damping of orthotropic beams is considered as a function of material orientation and the papers also consider cross-ply laminates and angle-ply laminates, as well as more general types of symmetric laminates. The damping concept of Adams and Bacon was also applied by Adams and Maheri [12] to the investigation of angle-ply laminates made of unidirectional glass fibre or carbon layers. The finite element analysis has been used by Lin et al [13] and by Maheri and Adams [14] to evaluate the damping properties of free-free fibre reinforced plates. These analyses were extended to a total of five damping parameters, including the two transverse shear damping parameters. More recently the analysis of Adams and Bacon was applied by Yim [9] and Yim and Jang [15] to different types of laminates, then extended by Yim and Gillespie [16] including the transverse shear effect in the case of 0° and 90° unidirectional laminates. For thin laminate structures the transverse shear effects can be neglected and the structure behaviour can be analysed using the classical laminate theory. The natural frequencies and mode shapes of rectangular plates are well described using the Ritz method introduced by Young [17] in the case of homogeneous plates. The Ritz method was applied by Berthelot and Sefrani [10] and Berthelot [18] to describe the damping properties of laminate beams and plates. The results derived from these analyses were first applied to the evaluation of damping parameters of materials from the flexural vibrations of beam specimens and compared to the experimental results. Next, damping of different laminates was considered. The purpose of this chapter is to report an extended synthesis of the recent developments on the evaluation of the damping of laminates and sandwich materials. Modelling of damping as well as experimental investigation will be considered. The different concepts introduced will be last applied to the analysis of the dynamic response of a simple shape damped composite structure.
2. Damping in a Unidirectional Composite as a Function of the Constituents The elastic behaviour of a unidirectional orthotropic material is characterized by the engineering constants
L, T , T , also
noted
EL , ET , LT and GLT ,
1, 2, 3 .
measured in the material directions
In the same way, the damping properties can be
described by four damping coefficients. In practice, damping associated to the Poisson’s ratio
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
55
is neglected, and the evaluations of the damping coefficients associated to the longitudinal and transverse Young’s moduli and to the shear modulus are generally based on an energy approach. The use of the energy approach to evaluate the damping properties of a structure was introduced by Ungar and Kerwin [19], considering that the structural damping can be described as a function of the constitutive elements of the structure and of the energy stored in these elements: n
iUi i 1 n
U i i 1
.
(1)
Applying this relation to a fibre composite leads to express the damping c of the composite as function of the fibre damping f and matrix damping m according to the expression:
c
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
where
U f , U m and U c
f U f mU m Uc
,
(2)
are the elastic energies stored in fibres, matrix and composite
material, respectively. Expression (2) is general, but in practice the application is restricted to simple fibre-matrix arrangements and loading conditions for which the elastic energies stored can be derived easily. Applying Expression (2) to the case of a unidirectional fibre composite loaded in the fibre direction leads to the expression of the longitudinal damping as:
L f
where
Vf
Ef E Vf m m 1 Vf EL EL ,
is the fibre volume fraction,
matrix, respectively, and
EL
Ef
and
Em
(3)
are the Young’s moduli of fibres and
is the Young’s modulus of the unidirectional composite. This
modulus is well evaluated by the law of mixtures and Expression (3) can be written as:
L
Vf 1 Vf f m Em Ef Vf 1 Vf 1 Vf Vf Ef Em .
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(4)
56
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
In the case where the damping of fibres can be neglected, Expression (4) is simply reduced to:
L
1 Vf m Ef 1 V V f f Em .
(5)
If now the unidirectional fibre composite is loaded in the transverse direction, Expression (2) leads to the transverse damping which can be expressed as:
T f
ET E Vf m T 1 Vf Ef Em ,
(6)
introducing the transverse Young’s modulus of composite. This modulus can be expressed by an inverse law of mixtures, but a better evaluation can be obtained [20, 21] using expression:
ET
where
KL
2 1 1 2 2 LT 2 K L 2GTT EL ,
(7)
is the lateral compression modulus of the unidirectional composite and
GTT
the
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
transverse shear modulus. These coefficients are deduced from the expressions established by Hashin [22, 23] and Hill [24]:
Vf
K L Km
1 kf k m
1 3
Gf Gm
1 Vf km 34 Gm
,
(8)
and by Christensen and Lo [25, 26]:
Vf GTT Gm 1 Gm K m 2Gm 1 Vf Gf Gm 2 K m Gm . The bulk moduli moduli
( K m , Kf )
(km , kf ),
the shear moduli
(Gm , Gf )
(9)
and the lateral compression
of the matrix and fibres are expressed as functions of the Young’s moduli
and Poisson’s ratios of the matrix and fibres by:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
ki
Ei , 3 1 2 i
Gi
Ei , 2 1 i
Ki ki
Gi , 3
57
i m, f. (10)
The Poisson’s ratio LT in Expression (7) can be evaluated by the law of mixtures. Lastly, in the case of a longitudinal shear loading, the expression of composite damping is similar to Expression (6) obtained in the case of a transverse loading:
LT f
GLT G Vf m LT 1 Vf Gf Gm ,
(11)
where the longitudinal shear modulus can be evaluated [24, 25] by:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
GLT Gm
Gf 1 Vf Gm 1 Vf Gf 1 Vf Gm 1 Vf
.
(12)
Limited experimental results are reported in literature on the processes of composite damping at the scale of fibres, matrix and fibre-matrix interface. Adams et al. [27] found that the longitudinal damping of unidirectional carbon-fibre composites and glass-fibre composites fell rapidly with increasing the fibre volume fraction. Both composites have essentially the same damping for a given volume fraction. It was found by Adams [28] that Expression (5) considerably underestimates the experimental values of the longitudinal damping. Several factors were thought to contribute to the discrepancy: fibre misalignment, imperfections in the materials (matrix cracks and fibre-matrix debonding), effect of fibre-matrix interface. Fibre interaction and fibre-matrix interphase were considered in [2831], in the case of discontinuous fibres. Authors estimate the strain energies stored in fibres and matrix using a finite element analysis. Then, composite damping was derived from Expression (2). More recently, Yim [32], and Yim and Gillepsie [16] have considered the evaluation of the damping parameters in the case of unidirectional carbon-fibre epoxy composites. According to the results obtained by Adams [28], Yim [32] introduced a curve fitting parameter in relation (5) and expressed the longitudinal damping as:
L
1 Vf
E 1 Vf f Vf Em
m .
(13)
In fact, the curve fitting parameter is obtained by Yim considering the only fibre fraction equal to 0.65. In the same way, parameters were introduced in expression (6) for the transverse damping and in Expression (11) for the longitudinal shear damping. Recently, Berthelot and Sefrani [33] consider the description of the longitudinal and transverse damping of unidirectional composites, introducing different damping coefficients associated with the motions of fibres in the longitudinal and transverse directions. A new
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
58
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
model is developed for describing the transverse damping of composites. The results are compared to the experimental results obtained for glass fibre, carbon fibre and Kevlar fibre composites.
3. Bending Vibrations of Undamped and Damped Laminate Beams 3.1. Undamped Beam Vibrations 3.1.1. Normal Modes in the Case of Undamped Vibrations The differential equation of motion for an undamped beam may be written [20, 21] as:
s
where
w0 w0 ( x, t )
2w 0 t 2
ks
4w 0 x 4
0, (14)
is the transverse displacement of the beam at point of coordinate x, s
is the mass per unit area and ks is the stiffness per unit area given by:
ks
1 , 1 D11
(15)
1
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The term D11 is the 11-component of the matrix inverse of the bending stiffness matrix. Equation (14) of transverse vibrations may be rewritten in the form:
2w 0 t 2
02 a 4
4w 0 x 4
0, (16)
introducing the natural angular frequency of the undamped beam:
0
1 ks 1 2 2 a s a
1 . 1 s D11
(17)
When the beam vibrates in its ith natural mode, the harmonic transverse displacement at a point of coordinate x is:
w0 x, t X i x A cos it B sin it ,
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(18)
Damping in Composite Materials and Structures
59
where Xi(x) is the normal shape of the natural mode and ωi is its angular frequency. Substitution of Equation (18) into Relation (16) results in:
d 4 X i 1 i2 X i 0. dx 4 a 4 02
(19)
The general solution for Equation (19) may be written as:
X i x Ci sin i
x x x x Di cos i Ei sinh i Fi cosh i , a a a a
(20)
where parameter i is given by:
i . 0
i
(21)
The parameter i and the constants Ci, Di, Ei and Fi in Equation (20) must be determined (to within an arbitrary constant) from the boundary conditions at the ends of the beam. Then, the normal modes can be superimposed to obtain the total response of the beam as:
w0 x, t X i x Ei cos it Fi sin it .
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
i 1
(22)
Orthogonality and normality properties of the functions Xi are considered in [20, 21, 34, 35].
3.1.2. Motion Equation in Normal Co-ordinates When a load q(x, t) is imposed, the motion equation (14) of a beam becomes:
s
2w 0 t 2
ks
4w 0 x 4
q ( x, t ), (23)
The transverse displacement w0(x, t) can be expressed in terms of time functions i(t) and normal displacement functions Xi(x) as:
w0 x, t i t X i x . i 1
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(24)
60
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
Substitution of this equation into the motion equation (23), then considering the orthogonality and normality properties of the normal functions leads to:
i i2i pi t ,
i 1,2,...,
(25)
where
pi t
a
0
p x, t X i dx, (26)
with
p x, t
1
s
q x, t . (27)
Equation (25) is the motion equation expressed in normal coordinates.
3.2. Damping Modelling Using Viscous Friction
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
3.2.1. Vibration Equation of Damped Beams Among all the sources of energy dissipation, the case of viscous damping where the damping force is proportional to velocity is the simplest to deal with mathematically. For this reason damping forces of a complicated nature are generally replaced by equivalent viscous damping. In this case, the damping force is proportional to the velocity. Thus, the differential equation of motion for a damped beam is deduced from Equation (23) and is written as:
s
2w 0 t 2
cs
w0 4w 0 ks q ( x, t ), t x4
(28)
introducing the coefficient of viscous damping cs by unit area. Then, Equation (28) can be rewritten in the following form:
2w0 t 2
cs w0 4w0 02 a 4 p( x, t ), s t x4
(29)
introducing the angular frequency (17) of the undamped beam and where the reduced load p( x, t ) is defined in Equation (27).
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
61
3.2.2. Motion Equation in Normal Coordinates As in the case of undamped beam (Section 3.1.1), the motion equation (29) can be transformed in an equation in normal coordinates by introducing the transverse displacement expressed by Equation (24). We obtain:
i 2i i i i2 i pi ,
i 1, 2, ... ,
,
(30)
introducing the modal damping coefficient i, related to the coefficient of viscous damping by:
cs
s
2i i . (31)
Each of the equations (30) is uncoupled from all the others, and the response i(t) of each mode i can be determined in the same manner as for one-degree system with viscous damping [36].
3.2.3. Forced Harmonic Vibrations In the case of a beam of length a, submitted to a harmonic load:
q( x, t ) qm ( x) cos t
,
(32)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
the component of the reduced load for the mode i is given by:
pi (t ) pm i ( x) cos t ,
(33)
with
pm i
1
s
a
qm X i dx. 0
(34)
Equation (30) of motion in normal coordinates becomes:
i 2i i i i2 i pm i cos t ,
i 1, 2,..., .
(35)
Considering the results obtained in the case of a system with one degree of freedom [36], the steady-state response for mode i is given by:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
62
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
i (t )
pm i
i2
Ki ( ) ai cos t bi sin t , (36)
with
ai 1
Ki ( )
2 , i2
bi 2i
, i
1
.
2
1 2 2i i i 2
(37)
2
(38)
Then, the transverse displacement is deduced from (24), which gives:
w 0 ( x, t ) i 1
pm i
i2
Ki ( ) X i ( x) ai cos t bi sin t . (39)
The equation of the harmonic motion can be expressed in the frequency domain in the complex form:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
i ( ) H i ( ) Pi ( )
i 1, 2,...,
(40)
where Φi(ω) and Pi(ω) are the complex amplitudes associated to the time functions i(t) and pi(t), respectively, and introducing the transfer function:
H i ( )
1
i2
H ri ( ), (41)
with
H ri ( )
1 . 2 1 2 2i i i i
(42)
Hri is the reduced transfer function. The time response i(t) in complex form is then deduced from Equation (40) and expressed in the form (39) with:
Ki ( ) H ri ( ) ,
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(43)
Damping in Composite Materials and Structures
63
and
ai Re H ri ,
bi Im H ri .
(44)
3.3. Damping Modelling Using Complex Stiffness As considered in the case of one degree system [36], the energy dissipation in the case of harmonic vibrations can be accounted for by introducing the complex stiffness per unit area:
ks* ks 1 i ,
(45)
where η is the structural damping coefficient or the loss factor introduced in Section 2. It results that motion equation (28) can be transposed in complex form using the following procedure:
i 1
2 s i
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
ks i
a
X i X j d x i ks i
0
a
d 4 Xi
0
dx 4
a
d 4 Xi
0
dx 4
X j dx
X j d x s Pj ( ),
(46)
This equation introduces the complex amplitudes Φi(ω), Xi(ω), Xj(ω), and Pj(ω) of i(t), xi(t), xj(t) and pj(t), respectively. Considering the orthogonality and normality relations, Equation (46) can be rewritten as:
i2 2 i i2 i i ( ) Pi ( ),
i 1, 2,..., , (47)
introducing the loss factor ηi of each mode. Equations (47) constitute the motion equation in normal coordinates. These equations are uncoupled. They can be written in form (40) with:
H ri ( )
1 i 2 1 2 ii
. (48)
Finally, the transverse displacement can be expressed in form (39), with
2 ai 1 2 , i
bi i ,
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(49)
64
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
and
1
Ki ( )
2
.
2 1 2 i i 2
(50)
3.4. Beam Response to a Concentrated Loading In the case of a load concentrated at point x = x1 of a beam, the exerted loading can be written as:
q( x, t ) q( x1, t ) ( x x1 ) q1(t ), where
(51)
( x x1 ) is the Dirac function localized at x1. The modal component of the reduced
load is:
pi (t ) p1(t )
a
X i ( x) ( x x1 ) d x,
0
(52)
which yields:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
pi (t ) X i ( x1 ) p1(t ),
(53)
with
p1(t )
1
s
q1(t ). (54)
In the case of an impact, the reduced load can be expressed as:
p1(t ) p1 (t ),
(55)
where p1 is constant and (t) is the impulse Dirac function localized at time t = 0. This function can be expanded in Fourier transform as:
(t )
eit d.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(56)
Damping in Composite Materials and Structures
65
Thus, the impact loading generates the whole frequency domain, and for every frequency the motion equation in normal coordinates is written in form (35) with:
pm i p1 X i ( x1 ).
(57)
Equation (36) can also be written in form (40) where the transfer function is expressed by (42) in the case of the damping modelling using viscous friction or by (48) in the case of modelling using complex stiffness. Thus, it results that the transverse displacement can be written as:
w 0( x, t ) p1
X ( x ) X ( x)
1
i
1
i
2 i
i 1
K i ( ) (ai cos t bi sin t ), (58)
where ai, bi and Ki are given by (37) and (38) in the case of the damping modelling using viscous friction and by (49) and (50) in the case of the damping modelling using complex stiffness.
4. Evaluation of the Damping Properties of Orthotropic Beams as Functions of The Material Orientation 4.1. Energy Analysis of Beam Damping
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
4.1.1. Introduction The prediction for damping properties of orthotropic beams as functions of material orientation was developed by Adams and Bacon [8] and Ni and Adams [11]. These analyses also consider cross-ply laminates and angle-ply laminates, as well as more general types of symmetric laminates. The damping concept of Adams and Bacon was also applied by Adams and Maheri [12] to the investigation of angle-ply laminates made of unidirectional glass fibre or carbon layers. More recently the analysis of Adams and Bacon was applied by Yim [9] and Yim and Jang [15] to different types of laminates, then extended by Yim and Gillespie [16] including the transverse shear effect in the case of 0° and 90° unidirectional laminates.
4.1.2. Adams-bacon Approach For an orthotropic material the strain-stress relationship in material axes (L, T, T') = (1, 2, 3) is given [20, 21] by:
1 S11 S 2 12 6 0
S12 S22 0
0 1 0 2 , S66 6
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(59)
66
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
where the components Sij are the compliance constants related to the engineering moduli EL, ET, GLT and LT by the following expressions:
S11
1 1 1 , S12 LT , S22 , S66 . EL EL ET GLT
(60)
Adams and Bacon [8] consider that the strain energy stored in a volume element V can be separated into three components associated respectively to the stresses 1, 2 and 6 expressed in the material axes as:
U U11 U 22 U 66 ,
(61)
with
1 2
1 2
1 2
1 2
(63)
1 2
1 2
(64)
U11 1 1 V 1 S111 S12 2 V , U 22 2 2 V 2 S12 1 S22 2 V ,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
U 66 6 6 V 62 S66 V .
(62)
Thus, Adams and Bacon consider that the energy U11 is the strain energy stored in tension-compression in the longitudinal direction, U22 is the strain energy stored in tensioncompression in the transverse direction and U66 is the strain energy stored in in-plane shear. Then, the strain energy dissipation in the longitudinal direction is written as:
U11 11 U11 ,
(65)
introducing the longitudinal specific damping capacity 11 measured in the case of tractioncompression tests of 0° materials and assuming the damping is independent of the applied stress 1. Expressions (62) and (65) yield:
1 2
U11 111 S111 S12 2 V . Similarly, the strain energy dissipation in the transverse direction is expressed as:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(66)
Damping in Composite Materials and Structures
1 2
U 22 22 2 S121 S22 2 V ,
67
(67)
introducing the transverse specific damping capacity 22. And the strain energy dissipation in shear deformation is given by:
1 2
U 66 66 62 S66 V ,
(68)
introducing the in-plane shear damping specific capacity 66. Hence, the total energy dissipated in the element can be written as:
U U11 U 22 U 66 .
(69)
This expression can be extended to the whole volume of the laminate to derive the total energy dissipation:
U
U , V
(70)
and the specific damping capacity of the laminate is then:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
U , U
(71)
with
U
The stresses
1, 2 and 6 ,
U V
(72)
expressed in the material directions are related [20, 21]
to the stresses xx , yy and xy , in the beam directions by the relation: 2 1 cos sin 2 2 6 sin cos
sin 2 cos 2 sin cos
2sin cos xx 2sin cos yy cos 2 sin 2 xy ,
where θ is the orientation of the orthotropic material with respect to the beam directions. Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(73)
68
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
In the case of free flexure of the beam along the x direction, the stresses yy and xy are zero, and the stresses in the material directions are:
1 xx cos 2 , 2 xx sin 2 , 6 xx sin cos .
(74)
The energy dissipated in an element of unit volume is given by:
1 2 U xx 11 S11 cos 2 S12 sin 2 cos 2 2
22 S12 cos 2 S22 sin 2 sin 2 66 S66 cos 2 sin 2 .
(75)
The strain energy stored in the element is:
1 1 2 , U xx xx xx S11 2 2
(76)
where
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
S11
1 1 1 1 cos4 sin 4 2 LT Ex EL ET EL GLT
2 2 cos sin ,
(77)
introducing the Young's modulus measured in the x direction [20, 21]. Thus, Relations (74) to (77) lead to the expression of the specific damping capacity in the x direction:
x Ex 11 cos 4 22 sin 4 66 11 22 LT cos 2 sin 2 . EL
ET
GLT
EL
(78)
4.1.3. Ni-Adams Analysis In this section the analysis of Ni-Adams [11] is developed in the particular case of the bending of a beam constituted of an orthotropic or unidirectional material. The beam of length a and width b is caused to vibrate along its length (the x direction). In the analysis, only the principal bending moment Mx is applied along the x direction, the other moments being zero: My = Mxy = 0, according to the assumptions of the classical laminate theory. Thus curvatures are expressed [20, 21] as:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
69
1 x D11 Mx, 1 y D12 Mx, 1 xy D16 M x.
(79)
1
where the Dij coefficients are the flexural compliance matrix components, derived as the elements of the matrix inverse of [Dij] expressed in the beam axes. The curvature x is due to bending along the x direction, the curvature y is due to the Poisson coupling and the curvature xy results from the bending-twisting coupling. In the case of beam bending, the strain field [20, 21] is reduced to:
xx z x , yy z y , xy z xy .
(80)
The stresses in the material, referred to the plate directions, are deduced from the stressstrain relation [20, 21] as:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Q12 Q11 xx Q xy z Q12 22 Q16 Q26 yy
x Q16 y . Q26 Q66 xy
(81)
are referred to the plate axes x and y, and are expressed [20, The reduced stiffnesses Qij 21] as functions of the reduced stiffnesses Qij in the material directions by the expressions reported in Table 1. Considering Equations (79) to (81) leads to:
1 1 1 D11 D12 D16 yy z Q12 Q22 Q26 M x, 1 1 1 D11 D12 D16 xy z Q16 Q26 Q66 M x. 1 1 1 D11 D12 D16 xx z Q11 Q12 Q16 M x,
(82)
Then, the stresses expressed in the material directions are deduced from Equation (73). As previously, Ni and Adams consider that, in the case of free bending beam, the stresses
yy and xy can be neglected. Thus, the stresses in material directions are given by:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
70
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
1 1 1 D11 D12 D16 2 z Q11 Q12 Q16 M x sin 2 , 1 1 1 D11 D12 D16 6 z Q11 Q12 Q16 M x sin cos . 1 1 1 D11 D12 D16 1 z Q11 Q12 Q16 M x cos 2 ,
(83)
The strains in the material directions can be expressed as functions of the strains in the beam directions considering the strain transformations [20, 21]. We obtain: 2 1 cos sin 2 2 6 2sin cos
sin 2 cos 2 2sin cos
xx sin cos yy , cos 2 sin 2 xy sin cos
(84)
Considering that yy is much smaller than xx and xy , the strain yy can be neglected, and the strains in the material directions are given by: Table 1. Reduced stiffness constants of a unidirectional or orthotropic layer, off its material directions
Q11 cos 4 Q22 sin 4 2 Q12 2Q66 sin 2 cos 2 , Q11
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Q11 Q22 4Q66 sin 2 cos 2 Q12 sin 4 cos 4 , Q12 Q11 Q12 2Q66 sin cos3 Q12 Q22 2Q66 sin 3 cos , Q16 Q11 sin 4 Q22 cos 4 2 Q12 2Q66 sin 2 cos 2 , Q22 Q11 Q12 2Q66 sin 3 cos Q12 Q22 2Q66 sin cos3 , Q26
Q11 Q22 2 Q12 Q66 sin 2 cos 2 Q66 sin 4 cos 4 . Q66
1 1 2 z D11 sin 2 D16 sin cos M x ,
1 1 1 z D11 cos 2 D16 sin cos M x ,
1 1 6 z 2 D11 sin cos D16 cos 2 sin 2 M x .
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(85)
Damping in Composite Materials and Structures
71
As in the Adams-Bacon approach, the energy dissipation is separated into three components associated with the stress components 1, 2 and 6 expressed in the material directions. Thus, the energy dissipation can be expressed as:
U U11 U 22 U66 ,
(86)
with
U11 b
U 22 b
U 66 b
a
x 0
2
a
x 0
2
a
x 0
2
h/2 z 0 h/2 z 0 h/2 z 0
1 d x d z, 2 11 1 1
(87)
1 d x d z, 2 22 2 2
(88)
1 d x d z. 2 66 6 6
(89)
These expressions lead to:
U11
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
U 22
1 D11 cos 2
1 D16 sin
0 M x2 d x, a
cos
(90)
1 1 1 1 D11 D12 D16 22 I Q11 Q12 Q16 sin 2 2
U 66
1 1 1 1 D11 D12 D16 I Q11 Q12 Q16 cos 2 2 11
1 D11 sin 2
1 D16 sin
cos
0 M x2 d x, a
(91)
1 1 1 1 D11 D12 D16 66 I Q11 Q12 Q16 sin cos 2 1 1 2 D11 sin cos D16 cos 2 sin 2
a 0
M x2 d x, (92)
introducing the quadratic moment I of the cross-section of the beam with respect to the (x, y) plane:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
72
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
I
b h3 , 12
(93)
where h is the beam thickness. The total strain energy of the beam can be expressed [20, 21] as:
U
1 2
a
b/2
x 0
y b / 2
M x x M y y M xy xy dx dy. (94)
The moments My and Mxy are neglected and the total strain energy can be expressed as:
b 1 U D11 2 Then, the specific damping capacity
fx
a x 0
M x2 d x. (95)
for the beam bending along the x-direction is
given by:
fx
U11 U 22 U 66 . U
(96)
In the case of a beam constituted of the same orthotropic or unidirectional material, the
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
stiffness constants Dij of the beam are related to the reduced stiffness constants Qij of the material by the expression:
Dij Qij
and the compliance components
Qij1
Qij .
(97)
Dij1 are given by: Dij1
where
h3 , 12
12 1 Qij , h3
are the components of the inverse matrix
(98)
Qij 1 of the reduced stiffness matrix
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
73
4.1.4. General Formulation of Damping Expressions obtained by the analysis of Adams-Bacon (78), then by the analysis of NiAdams (96) show that the specific damping capacity evaluated in the direction can be expressed in the general form:
11 a11 22 a22 66 a66 .
(99)
Functions aij( ) differ according to the analysis which is considered. In the case of Adams-Bacon approach, functions aij( ) are expressed as:
a11
1 S11 cos 2 S12 sin 2 cos 2 , S11
1 S12 cos 2 S22 sin 2 sin 2 , S11 1 a66 S sin 2 cos 2 . 66 S11 a22
(100)
In the case of the analysis of Ni-Adams, functions aij( ) are given by:
1 1 Q11 cos 2 Q16 sin cos cos 2 , 1 1 1 1 Q11 Q12 Q16 a22 1 Q11 Q12 Q16 Q11 1 1 Q11 sin 2 Q16 sin cos sin 2 , 1 1 1 1 Q11 Q12 Q16 a66 1 Q11 Q12 Q16 Q
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
a11
1 1 1 1 Q11 Q12 Q16 Q11 Q12 Q16 1 Q11
11
1 2 Q11 sin cos Q161 cos 2 sin 2 sin cos .
(101)
4.2. Complex Moduli The correspondence principle (Section 1) can be applied to the effective bending modulus of a beam [20, 21]. In complex form this bending modulus is expressed as:
E *fx
12 * 1 h D11 3
,
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(102)
74
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
where
* 1 D11 is
expressed as a function of the complex moduli of the laminated material.
Relation (102) allows us to evaluate the loss factor E associated to the bending modulus as: fx
E *fx E fx 1 i E
fx
.
(103)
This complex modulus has been also considered by Yim and Jang [15]. Previous relations correspond to the case of the free flexure of laminate beam where Mx is the only applied moment, curvatures being expressed by (79). Adams and Bacon [8] also consider the case of a pure flexure for which the twisting would be constrained to zero xy = 0. Considering the curvature-moment relations [20, 21], this pure flexure would be obtained when the twisting moment would be equal to:
M xy
1 D16
* D66
Mx, (104)
and the curvature-moment relations yield:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
2 1 D16 M . 1 x D11 x 1 D66
(105)
This expression is substituted for Expression (79) of x obtained in the case of free flexure and expressions of the effective bending modulus becomes:
1
E fpx 1
1 D16
2
E fx .
1 1 D11 D66
(106)
In fact, the scheme of pure flexure is theoretic, since there exists a bending-twisting coupling for off-axis materials. Moreover E fpx E fx . However this scheme was considered by Yim and Jang [15] and applied to the damping of beam flexure introducing the complex bending modulus:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
1
E*fpx 1
D* 1 16
2
75
E*fx .
* 1 * 1 D11 D66
(107)
5. Evaluation of the Damping Properties of Plates as Function of Material Direction 5.1. Orthotropic Plates 5.1.1. Formulation The energy approach considered in the previous section for the damping of beams can also be applied for evaluating the damping properties of plates. The energy approach is based on the evaluation of the strain energy, which can be derived by finite element analysis in the case of a structure of complex shape or by using the Ritz method in the case of the analysis of rectangular plates. This analysis has been developed in [10, 18] and is considered hereafter. In the Ritz method [20, 21], the transverse displacement is expressed in the form of a double series of the coordinates x and y: M
w 0 ( x, y )
N
Amn X m ( x) Yn ( y)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
m 1 n 1
,
(108)
where the functions Xm(x) and Yn(y) have to form a functional basis and are chosen to satisfy the essential boundary conditions along the edges x = 0, x = a and y = 0, y = b. The coefficients Amn are next determined from the stationarity conditions which make extremum the energy function:
U d max Ec max 0, Amn where
m 1, 2, . . . , M , n 1, 2, . . . , N ,
(109)
U d max Ec max is the energy obtained by substituting Expression (108) for the
transverse displacement into the expression of the energy function:
U d max Ec max
1 2
a
b
x 0
y 0
2 2 2 2 2 2 D11 w 0 2 D12 w 0 w 0 D22 w 0 x 2 y 2 x 2 y 2
2w 0 2 2 4 D66 s w 0 d x d y. xy 2
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(110)
76
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
U d max
and
Ecmax are
the maximum strain energy and maximum kinetic energy,
respectively, during a cycle of harmonic plate vibrations. The strain energy Ud can be expressed as a function of the strain energies related to the material directions as:
U d U1 U 2 U 6 ,
(111)
with
dx dy dz, 1 d x d y d z, 2 1 d x d y d z, 2
U1 U2 U6
1 2
1 1
2 2
6 6
(112)
where the triple integrations are extended over the volume of the plate. Considering the case of a plate constituted of a single layer of unidirectional or
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
orthotropic material, the strains 1, 2 and 6 are related to the strains xx, yy and xy in the beam directions according to the strain transformations. The strain transformations are obtained inverting Expression (84): 2 xx cos 2 yy sin xy 2sin cos
sin 2 cos 2 2sin cos
sin cos 1 sin cos 2 , 2 2 cos sin 6
(113)
Next the stresses 1, 2 and 6 can be evaluated considering the elasticity relations of plates:
1 Q11 1 Q12 2 , 2 Q12 1 Q22 2 , 6 Q66 6 .
(114)
It results that the strain energy U1, stored in tension-compression in the fibre direction, can be written as:
U1 U11 U12 , Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(115)
Damping in Composite Materials and Structures
77
with
Q 1 Q 2
U11 U12
1 2
2 11 1
dx dy dz,
12 1 2
d x d y d z. (116)
Expression (115) separates the energy U11 stored in the fibre direction and the coupling energy U12 induced by the Poisson’s effect. They are given by:
U11
1 2
Q
11
xx2 cos4 yy2 sin 4 xy2 sin 2 cos2
2 xx yy sin 2 cos 2 2 xx xy sin cos3
2 yy xy sin 3 cos d x d y d z, (117)
U12
1 2
Q
12
2 2 2 xx sin 2 cos 2 yy sin 2 cos 2 xy sin 2 cos 2
xx yy sin 4 cos 4 xx xy sin 2 cos 2 sin cos (118)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
yy xy cos2 sin 2 sin cos d x d y d z. In the case of bending vibrations of plates, the strains are deduced from Equations (80), which lead to the relations with the transverse displacement:
xx z yy z
2w 0 x2 2w 0
xy 2 z
y 2
, ,
2w 0 . x y
(119)
Then, the strain energies U11 and U12 are expressed as functions of the transverse displacement introducing expressions (119) in Equations (117) and (118), respectively. Next, considering the Ritz method, the transverse displacement is introduced in the form (108) and the expressions of the energies are integrated over the plate volume. Calculation leads to the following correspondences considered in [20, 21]:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
78
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. 2 2200 xx Cminj ,
2 xx yy
2 0022 4 yy Cminj R ,
2 2 xy 4 C1111 minj R ,
1 2002 0220 Cminj Cminj R2 , 2
1012 0121 2 yy xy Cminj Cminj R3 , 1210 2101 2 xx xy Cminj Cminj R,
(120)
pqrs
where the coefficients Cminj are expressed as: pqrs pq rs Cminj I mi J nj ,
(121)
introducing the dimensionless integrals: pq I mi
rs J nj
1
d p X m dq X i
0
du
1 r
p
du
q
s d Yn d Y j
0
dv
r
dv
s
m, i 1, 2...M , p, q 0,1, 2,
d u,
(122)
n, j 1, 2...N ,
dv ,
r , s 0,1, 2.
(123)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
pq rs The integrals I mi and J nj are calculated using the reduced coordinates:
u x / a,
v y / b,
and
(124)
where a and b are the length and the width of the plate, respectively. It results that the strain energies U11 and U12 can be written in the form:
U11
1 2R a2
M
N
M
N
Amn Aij D11 f11 , m1 n1 i 1 j 1
(125)
with
2200 0022 4 2002 2 2 2 f11 Cminj cos4 Cminj R sin 4 2 2 C1111 minj Cminj R sin cos 2101 0121 3 4Cminj R sin cos3 4Cminj R sin 3 cos ,
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(126)
Damping in Composite Materials and Structures
D11 Q11
h3 , 12
79
(127)
and
U12
1 2R a2
M
N
M
N
Amn Aij D12 f12 , m1 n1 i 1 j 1
(128)
with
2200 0022 4 2 2 2 2002 2 4 4 f12 Cminj Cminj R 4C1111 minj R sin cos Cminj R cos sin
2101 0121 3 2 Cminj R Cminj R sin 2 cos2 sin cos ,
D12 Q12
h3 . 12
(129)
(130)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
These expressions introduce the length-to-width ratio of the plate (R = a/b). In the same way, the energy U2 stored in tension-compression in the direction transverse to the fibre direction is obtained as:
U 2 U 21 U 22 ,
(131)
U 21 U12 ,
(132)
with
and
U 22
1 2R a2
M
N
M
N
Amn Aij D22 f22 , m1 n1 i 1 j 1
(133)
with
2200 0022 4 2002 2 2 2 f 22 Cminj sin 4 Cminj R cos4 2 2 C1111 minj Cminj R sin cos 2101 4Cminj R sin 3
cos
0121 3 4Cminj R sin
D22 Q22
cos ,
h3 . 12
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(134)
3
(135)
80
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. Lastly, the strain energy U66 stored in in-plane shear can be written as:
U 66
1 2R a2
M
N
M
N
Amn Aij D66 f66 , m1 n1 i 1 j 1
(136)
with
2200 0022 4 2002 2 f66 4 Cminj Cminj R 2Cminj R sin 2 cos 2 1111 2 4 Cminj R
0121 3 2101 cos2 sin 2 2 8 Cminj R Cminj R cos 2 sin 2 sin cos ,
D66 Q66
h3 . 12
(137)
(138)
Then, the energy dissipated by damping in the material is written in the form:
U 11 U11 212 U12 22 U 22 66 U66 ,
(139)
introducing the damping coefficients 11, 12, 22 and 66 associated to the strain energies, respectively. The strain energy U12 is generally negative, due to the coupling between 1 and
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
2, and the corresponding dissipated energy must be taken positive. In fact, this energy can be neglected with regard to the other energies. Next, the damping x in the x direction of the plate along its length is evaluated by the relation:
x
U . U
(140)
5.1.2. Procedure In the Ritz method, the functions Xm(x) and Yn(y) introduced in Expression (108) of the transverse displacement can be chosen [20, 21] as polynomials or as beam functions which give the characteristic shapes of the natural vibrations of beams (Section 3.1.1). The beam functions satisfy orthogonality relations which make zero many of the integrals (122) and (123). Functions Xm(x) and Yn(y) depend on the boundary conditions imposed along the plate edges [20, 21]. Integrals (122) and (123) can be next calculated by an analytical development or by a numerical process and stored. Then, the values of the integrals allow us to establish [36] the system of homogeneous equations for the undamped flexural vibrations of the plates. This system can be solved as an eigenvalue and eigenvector problem where the eigenvectors determine the vibration modes, whence the coefficients Amn for the transverse displacement
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
81
(110) corresponding to the different modes. Next, the different strain energies are derived, for a given mode, by reporting the values of the coefficients Amn in the energy expressions (125), (128), (133) and (136). Hence the laminate damping is derived from relation (140).
5.2. Laminated Plates The Ritz method used in the previous section for analyzing the damping properties of orthotropic plates can be also applied to arbitrary laminated plates [18]. In the present section we consider the case of a laminated plate constituted of n orthotropic layers (Figure 1). Each layer is referred to by the z coordinates of its lower face
hk 1 and upper face hk .
Layer can
also be characterized by introducing the thickness ek and the z coordinate zk of the middle plane of the layer. Layer orientation is defined by the angle
k
of layer axes with the axes (x,
y) of the plate. For a laminate, the strain energy relation (111) considered for a single orthotropic layer can be written in the axes of each layer as:
U dk U1k U 2k U 6k ,
(141)
and the total energy of laminate is given by: n
Ud
U1k U 2k U 6k
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
k 1
.
(142)
In the case of the vibrations of a rectangular plate of length a and width b, the strain energies are expressed by:
U1k
U 2k
U 6k
a
b
hk
x0
y 0
z hk 1
a
b
hk
x 0
y 0
a
b
x 0
y 0
z hk 1
1 1 dx dy dz, (143)
2 2 dx dy dz,
hk z hk 1
(144)
6 6 dx dy dz.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(145)
82
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. z layer number
n k
hk hk 1 1 h2 h 1 h 0
middle plane
2 1
Figure 1. Laminate element.
As in the previous subsection, the strain energy can be written in the form: n
Upqk ,
Ud
k 1 pq
(146)
with
1 2
U kpq
a
b
x 0
y 0
hk z hk 1
Q kpq kp qk dx dy dz,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
pq 11, 12, 22, 66.
(147)
By considering the Ritz method, the transposition of the results obtained previously in the case of a single layer leads to:
U kpq
1 2 Ra 2
Amn Aij f pq k h m 1 n 1 i 1 j 1 M
N
M
hk
N
k Q pq z 2 dz
k 1
.
(148)
Hence: k U pq
1 2 Ra 2
M
N
M
N
Amn Aij Dpqk f pqk ( ), m1 n1 i 1 j 1
(149)
with
D kpq
e3 1 3 hk hk31 Q kpq ek zk2 k Q kpq . 3 12
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(150)
Damping in Composite Materials and Structures
83
Then, the total energy dissipated by damping in the laminated plate is expressed as: n
U
11k U11k 212k U12k 22k U 22k 66k U66k , k 1
(151)
introducing the specific damping coefficient
k pq of each layer. Next, the damping x
in
the x direction of the plate along its length is evaluated by relation:
x
U U ,
(152)
where the dissipated energy is given by relation (151) and the total strain energy by relation (142). The functions f pq ( ) of each layer are simply derived from the functions f pq ( ) k
expressed previously in the case of a single layer of orthotropic material as: k k f pq ( ) f pq ( k )
,
(153)
where functions f pq ( ) are given by (126), (129), (134) and (137).
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
5.3. Conclusion The process for evaluating the laminate damping from the dissipated energy has been implemented by using the Ritz method. This procedure can also be carried out using a vibration analysis by the finite element method. In this case it is necessary to have access to the strain and stress fields for each vibration mode. Next, the energies and the loss damping are obtained in the same way as for the Ritz method by considering the stored energies and the dissipated energies. Analysis by the finite element method will be considered in Section 6. The interest of the Ritz method lies in the fact that the process can be easily implemented with usual tools. However, the method is restricted to the analysis of beams or rectangular plates. In contrast, the finite element analysis can be applied to the case of a laminated structure of complex shape (Section 7).
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
84
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
6. Damping Analysis of Laminates with Interleaved Viscoelastic Layers
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
6.1. Introduction Constrained damping layers in isotropic metallic materials have been investigated in literature and the results obtained show that the layers provide significant higher damping than the initial materials. In the same way, inserting viscoelastic layers in laminates improves significantly the damped dynamic properties of the laminates. Moreover, the interlaminar damping concept is highly compatible with the fabrication processes of laminated structures. Limited analytical and experimental papers on the analysis of composite damping with viscoelastic layers have been reported in literature [37-42]. Saravanos and Pereira [37] develop a discrete-layer laminate theory for analysing the damping of composite laminates with interlaminar damping layers. Experimentally measured and predicted dynamic responses of graphite epoxy plates with co-cured damping layers are compared to illustrate the accuracy of the theory. Liao et al. [38] analyse the vibration-damping behaviour of unidirectional and symmetric angle-ply laminates as well as their interleaved counterparts with a layer of PEAA (polyethylene-co-acrylic acid) at the mid-plane. The introduction of the PEAA layer significantly improves the damping capability of laminates. The experimental results are compared with the results obtained by extending to laminate materials the evaluation of damping performances derived by Liao and Hsu [39] in the case of conventional constrainedlayer configuration: two isotropic outer layers and a thin viscoelastic interlayer. Shen [40] proposes an hybrid damping design which consists of a viscoelastic layer sandwiched between piezoelectric constraining cover sheets. The active damping component produces significant and adjustable damping, when the passive component increases gain. A first order shear deformation theory is used by Cupial and Niziol [41] to evaluate the natural frequencies and loss factors of a rectangular three-layered plate with a viscoelastic core layer and laminated faces. Simplified forms are discussed in the case of symmetric plate and for especially orthotropic faces. Comparison is made between the present shear deformation theory and simplified models. More recently the damping behaviour of a 0° laminated sandwich composite beam inserted with a viscoelastic layer was investigated by Yim et al. [42]. It is shown that the Ni-Adams theory [11] for evaluating the damping of laminate beams can be extended to evaluate the damping characteristics of laminated sandwich composite beams. Results show the capability of laminated sandwich composites with embedded viscoelastic layer to significantly enhance laminate damping. A finite element for predicting modal damping of thick composite and sandwich beams was developed by Plagianakos and Saravanos [43]. Previous linear layerwise formulations [44, 45] provided the basis for developing a discrete-layer higher order theory satisfying compatibility in interlaminar shear stress and modal damping was calculated by modal strain energy dissipation method. The effect of ply orientation of composite beams with interply viscoelastic damping layers was investigated. Experimental investigation of modal damping illustrated the accuracy of the developed formulation. The purpose of this Section is to show how the analysis of laminate damping developed in Section 5 can be extended to the case of the damping analysis of rectangular laminated
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
85
plates with interleaved viscoelastic layers. Modelling was developed in [46] and experimental investigation was implemented in [47].
6.2. Laminate Configurations Two types of laminates with viscoelastic layers were considered in [46, 47]: laminates with a single viscoelastic layer of thickness e0 interleaved in the middle plane of laminates (Figure 2) and laminates with two viscoelastic layers of thickness e0 interleaved away from the middle plane (Figure 3). The layers of the initial laminates are constituted of unidirectional or orthotropic materials with material directions making an angle θ with the x direction oriented along the length of plates under consideration. The total thickness of the unidirectional or orthotropic layers is e and the interlaminar layers are assumed to have an isotropic behaviour.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
6.3. Evaluation of the Damping in the Case of Interleaved Viscoelastic Layers The analysis developed in Sections 5.1 and 5.2 can be applied to the evaluation of the damping of laminates with interleaved viscoelatic layers. The results obtained for the in-plane damping shows that the analysis does not describe the experimental results obtained for damping in the case where one or two viscoelastic layers are interleaved. Indeed the in-plane energy stored in the viscoelastic layers is too low. This observation shows that the energy dissipation is induced by an other process, which leads to consider the transverse shear effects induced in the viscoelastic layers. The classical laminate theory which is considered in the previous analysis does not take account of the transverse shear effects induced in laminates. However, the classical laminate theory can be used to evaluate the transverse shear stresses in the different layers, in the following way. The in-plane stresses in the orthotropic layers are given by the relations:
xx yy xy
ort
Q12 Q16 xx Q11 Q22 Q26 yy , Q12 Q26 Q66 xy Q16
(154)
where the elements Qij are the reduced stiffness constants of the materials expressed in the (x, y) directions of the plate, which are deduced from the reduced stiffness constants Qij in the material directions, according to relations reported in Table 1. In the same way, the in-plane stresses in the viscoelastic layer are written as:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
86
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
e2
3
orthotropic layer
e0
2
viscoelastic layer
e2
1
orthotropic layer
h 2
Figure 2. Laminate with a single viscoelastic layer. z
e0 d
d
e 2
orthotropic layer
4
viscoelastic layer
h 2
d1
3
orthotropic layer
2
viscoelastic layer
1
orthotropic layer
e 2
e0
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
5
middle plane
d1
h 2
Figure 3. Laminate with two viscoelastic layers interleaved at the same distance from the middle plane. v v v Q11 Q12 xx v v yy Q12 Q22 xy 0 0
where the reduced stiffness constants
0 xx 0 yy , v Q66 xy
(155)
v Q pq are expressed as:
E 2 1 v E Q pq 1 2 0
E 1 2 E 1 2 0
0 , E 2 1 0
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(156)
Damping in Composite Materials and Structures
87
by introducing the Young’s modulus E and the Poisson’s ratio ν of the viscoelastic layer. The in-plane strains are expressed as functions of the transverse displacement relations (119). Thus, the in-plane stresses in the orthotropic layers are written as: ort xx
2w 0 2w 0 2w 0 z Q11 Q12 2 Q16 , x y x 2 y 2
2w 0
x 2
2w 0
x 2
ort yy z Q12
ort xy z Q16
Q22
Q26
2w 0 y 2 2w 0 y 2
2 Q26
2 Q66
(157)
2w 0 , x y
(158)
2w 0 , x y
(159)
and the in-plane stresses in the viscoelastic layer are: v xx
2 v 2w 0 v w0 z Q11 Q12 , x 2 y 2
2w 0
x 2
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
v v yy z Q12
v Q22
v v xy 2 z Q66
2w 0 , y 2
2w 0 . x y
(160)
(161)
(162)
The classical laminate theory neglects the transverse shear effects. However, the transverse shear stresses in the laminate layers can be derived from the fundamental equations of motion which can be expressed, neglecting the inertia terms, in the following forms: i i i xy xx xz 0, x y z i xy
x
iyy y
iyz z
0,
i ort, v, (163)
i ort, v.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(164)
88
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. The first equation (163) leads to: i i i xy xz xx z x y
i ort, v, (165)
which yields for the unidirectional or orthotropic layers: ort xz ort Axz x, y z , z
(166)
with ort Axz
Q11
3w0 x3
2Q66 Q12
3w0 x y 2
3 Q16
3w0 x 2 y
Q26
3w0 y3
. (167)
Integrating Relation (166), the transverse shear stress in the unidirectional or orthotropic layers is written as: ort xz
1 ort Axz x, y z 2 Cort . 2
(168)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Similarly, the transverse shear in the viscoelastic layer is given by: v xz
1 v Axz x, y z 2 Cv . 2
(169)
with v v Axz Q11
The constants
Cort
and
Cv
3w0 x3
v v Q12 2 Q66
3w0
x y2 .
(170)
in each layer are determined by considering the continuity of
the transverse shear stress at the interfaces between the viscoelastic layer and the orthotropic layers and that the transverse shear stress vanishes on the lower and upper faces of the laminate. The strain energy stored in xz-transverse shear by volume unit can be evaluated by the relation:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
u ixz
i 2 1 xz , i 2 Gxz
89
i ort, v, (171)
i
in which Gxz is the xz-transverse shear modulus of the orthotropic or viscoelastic layers. The total strain energy stored in the layers is next obtained by integration of Expression (171) over the whole volume of the plate. A similar development can be implemented, from Equation (166) for the evaluation of the transverse shear stresses
yz
and the strain energy stored in yz-transverse shear.
Finally, the total strain energy stored in the laminate with a viscoelastic layer can be written as: ort ort v v U U port U pv U xz U yz U xz U yz .
where the strain energy
(172)
U port and U pv are the in-plane strain energies stored in the ort
ort
v
orthotropic and viscoelastic layers, and U xz , U yz , U xz and energies stored in the orthotropic and viscoelastic layers . The strain energy can be written as:
v U yz are the transverse strain
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
ort ort ort ort U port U11 2U12 U 22 U 66 .
(173)
The specific damping coefficient ψ(θ) of the laminate can thus be evaluated by the relation: p s ( ) ort vp ort vs ,
(174)
where p ort
1 ort ort ort ort 11U11 212U12 22U 22 66U 66 U ,
vp v
U vp U ,
s ort ort ort ort xz U xz ort yz U yz
(175)
(176)
U1 ,
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(177)
90
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
vs
The specific damping coefficients
v p ort
v v U xz U yz
U and
. (178)
vp are the in-plane damping coefficients
considered in Relation (151). Expressions (174) to (177) introduce the specific damping coefficients xz and yz characterising the transverse shear energy dissipated in the unidirectional or orthotropic layers. For unidirectional materials these coefficients can be assimilated with the in-plane shear coefficient: ort
ort
ort xz ort yz 66
.
(179)
In the procedure used for the evaluation of damping, the stored strain energies are obtained using the Ritz method in a similar way as the one considered in Section 5. An extended development of this analysis is developed in [36, 46]. Moreover, the analysis shows how the modelling considered can be applied to the case of interleaved angle-ply laminates and to the case of laminates with external viscoelastic layers.
7. Damping Evaluation Using Finite Element Analysis
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
7.1. Introduction The Ritz method is restricted to the analysis of rectangular plates and to the case where the materials are unidirectional or orthotropic materials. In the case of other types of materials or complex shape structures, it is necessary to use the finite element method to analyse the dynamic behaviour.
z
k 1
lk 1 uk
k
k 1
lk u k 1
Figure 4. Stresses evaluated by finite element analysis in the layers of a finite element.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
91
Principle of finite element analysis of a dynamic problem of a structure with damping included is considered in different text-books. A synthesis is given in [36]. The energy approach for the evaluation of damping which has been developed in Section 5 by considering the Ritz method can be extended to any type of materials and to a complex shape structure by using a finite element analysis. The formulation is developed in the present section.
7.2. In-Plane Strain Energy as a Function of In-Plane Stresses When finite element based on the laminate theory with transverse shear effects included is used [36], finite element analysis gives, for a given mode of vibration, the values of stresses
xx , yy , xy , yz , xz ,
on the lower face (l) and upper face (u) of each layer k of each
finite element e of structure (Figure 4):
xxlk , yylk , xylk , yz lk , xz lk , xxuk , yyuk , xyuk , yz uk , xz uk .
(180)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
So, it is necessary to express the strain energy as a function of stresses. The in-plane strain energy for a given finite element e can be expressed as functions of the strain energies stored in the material directions according to Relations (111) and (112) introduced in Section 5 as:
Ude U1e U 2e U6e ,
(181)
dx dy dz, 1 d x d y d z, 2 1 d x d y d z, 2
(182)
with
U1e U 2e
1 2
1 1
e
2 2
e
U 6e
6 6
e
where the integration is extended over the volume of the finite element e. The in-plane strains 1,
2 and 6 related to the directions of the material of layer k are expressed as functions of stresses 1, 2 and 6 in the material directions according to the elasticity relation [20, 21] as:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
92
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
1 S111 S12 2 , 2 S121 S22 2 , 6 S66 6 ,
(183)
where the compliance constants of the layer k are:
S11
1 1 1 1 1 1 , S22 , S12 12 LT , S66 , E1 EL E2 ET E1 EL G12 GLT
(184)
introducing the engineering constants of layer material. e
It results that the strain energy U1 , stored in tension-compression in the L direction of layers can be expressed as: e e U1e U11 U12 ,
(185)
with
S 1 S 2
e U11 e U12
1 2
2 11 1
d x d y d z,
e
12 1 2
d x d y d z.
e
In each layer k, stresses
(186)
1, 2 and 6 , related to the material directions of the layer,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
can be expressed as functions of the in-plane stresses
xx , yy and xy ,
related to the
finite element directions (x, y, z) according to the stress transformations [20, 21]: 2 1 cos sin 2 2 6 sin cos
sin 2 cos 2 sin cos
2sin cos xx 2sin cos yy cos 2 sin 2 xy ,
(187)
where is the orientation of the material in the layer. Whence: e U11
1 2
e
2 2 S11 xx cos 4 yy sin 4
2 2 2 xy xx yy sin 2 cos 2
4xx xy sin cos3 4yy xy sin 3 cos dx dy dz,
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(188)
Damping in Composite Materials and Structures e U12
1 2
S
12
e
93
2 2 2 xx yy 4 xy sin 2 cos 2
xx yy sin 4 cos 4
2 yy xy yy xy
cos 2 sin 2 sin cos dx dy dz.
(189)
e
In the same way, the strain energy U 2 , stored in tension-compression in the T direction of each layer is obtained as: e e U 2e U 22 U12 ,
(190)
with e U 22
1 2
e
2 2 S22 xx sin 4 yy cos 4
2 2 2 xy xx yy sin 2 cos 2
4xx xy sin 3 cos 4yy xy sin cos3 dx dy dz,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
e U 6e U 66
1 2
S e
66
(191)
2 2 xx yy 2 xx yy sin 2 cos 2
2 4 xy cos 2 sin 2 2 yy xy xx xy cos 2 sin 2 sin cos d x d y d z. 2
(192)
Considering Expressions (188) to (192), the in-plane energies can be expressed as: n
e U11
k 1
where
n
e U11 k,
e U 22
n
e U 22 k,
k 1
e U12
n
e U12 k,
k 1
e U 66
U 66e k , k 1
(193)
e Upqk ( pq 11, 22, 12, 66) are the in-plane energies stored in layer k of the
element e and n is the number of layers. Introducing the terms:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
94
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
S 2 S 2
e U xxxxk
e U xyxyk
e U xxxyk
Se 2 e
e
hk
2 xxk d z,
hk 1 hk
2 xyk d z,
e U xxyyk
hk 1 hk
S 2 S 2
e U yyyyk
xxk xyk d z,
e U yyxyk
hk 1
Se 2 e
e
hk
2 yyk d z,
hk 1 hk
xxk yyk d z,
hk 1 hk
yyk xyk d z,
hk 1
(194)
where Se is the area of the finite element e. The in-plane energies stored in layer k of the element e can be expressed as: e e 4 e 4 U11 k S11k Uxxxxk cos k Uyyyyk sin k
e e 2 2Uxyxyk Uxxyyk sin 2 k cos 2 k e 4Uxxe xyk sin k cos3 k 4U yyxyk sin 3 k cos k ,
(195)
e e e 2 2 e U12 k S12 k Uxxxxk U yyyyk 4Uxyxyk sin k cos k e sin 4 k cos4 k Uxxyyk
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
e e 2 Uxxxyk Uyyxyk
sin 2 k cos2 k sin k cos k ,
(196)
e e 4 e 4 U 22 k S 22 k Uxxxxk sin k U yyyyk cos k
e e 2 2Uxyxyk Uxxyyk sin 2 k cos 2 k e 4Uxxe xyk sin 3 k cos k 4U yyxyk sin k cos3 k ,
(197)
e e e 2 2 e U 66 k S66 k Uxxxxk U yyyyk 2Uxxyyk sin k cos k 2 e cos2 k sin 2 k Uxyxyk
e e 2 Uxxxyk U yyxyk
sin 2 k cos2 k sin k cos k ,
These expressions introduce the orientation
k
of layer k.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(198)
Damping in Composite Materials and Structures
95
z
puk xxuk , yyuk , xyuk
hk ek
k
plk xxlk , yylk , xylk
hk 1
Figure 5. Stresses on the lower and upper faces of layer k.
7.3. In-Plane Stress Evaluation The laminate theory taking into account the transverse shear effects is based on a first order theory which expresses [20, 21] the displacement field as a linear function of z coordinate through the thickness of laminate element. It results that the in-plane stresses in layer k are linear functions of z coordinate of the forms:
pk apk ( x, y) z bpk ( x, y),
p xx, yy, xy.
,
(199)
with
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
e e pk xxk , eyyk , xyk .
(200)
Coefficients a pk and bpk in each element e can be deduced from the stresses calculated by the finite element analysis on the lower and upper faces of each layer k (Figure 5). Note that in-plane stresses are discontinuous at the layer interfaces. We obtain:
a pk
p u k p lk ek
,
b pk p uk p uk p lk
hk , ek
(201)
with
puk xxuk , yyuk , xyuk , plk xxlk , yylk , xy lk . and where ek is the thickness of the layer k and
hk
(202)
is the z coordinate of the upper face.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
96
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
7.4. In-Plane Energy Evaluation The energy terms
e e e Uxxxxk , Uyyyyk and Uxyxyk introduced by Equations (194) can be
expressed in the form: e Uppk
Se e I , 2 ppk
p xx, yy, xy, (203)
introducing the integral: e I ppk
hk
h
k 1
2 pk d z,
p xx, yy, xy. (204)
Considering Expression (199) of the in-plane stresses, this integral can be expressed as:
e I ppk
The energy terms
a 2pk 3 3 2 hk hk 1 a pk b pk hk2 hk21 b pk ek , 3 p xx, yy, xy, e e e Uxxyyk , Uxxxyk and Uyyxyk in Equations (194) are given by:
Se e I , 2 pqk
e U pqk Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
(205)
p, q xx, yy, xy, p q.
(206)
introducing the integral: e I ppk
hk
h
k 1
pk qk d z,
p xx, yy, xy. p q.
(207)
Introducing Expression (199) of the in-plane stresses into (207), we obtain:
1 1 e I pqk apk aqk hk3 hk31 apk bqk aqk bpk hk2 hk21 bpk bqk ek , 3 2 p, q xx, yy, xy, p q. (208) e e e e Finally, the strain energies U11k , U12k , U 22k and U66k stored in the layer k of the element e are given by Expressions (195) to (198) with:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
Se e I , 2 xxxxk S e e Ixx xyk , 2 S e e Ixx xyk , 2
Se e I , 2 yyyyk S e e I xxyyk , 2 S e e I yyxyk . 2
e U xxxxk
e U yyyyk
e U xyxyk
e Uxxyyk
e U xxxyk
e U yyxyk
97
(209)
The integrals Ipp and Ipq (p, q = xx, yy, xy) are expressed by Equations (205) and (208) with:
p u k p lk , apk ek hk b pk p u k p u k p l k e , k
q u k q lk , aqk ek b hk , qk q uk q uk q lk e k
p, q xx, yy, xy.
(210)
Next, the in-plane strain energies stored in element e are given by Expressions (193) and the total in-plane strain energies stored in the finite element assemblage is then obtained by summation on the elements as:
U11
e U11 ,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
elements
U 22
elements
U12
e U12 ,
elements e U 22 ,
U 66
elements
e U 66 .
(211)
7.5. Transverse Shear Stresses In the case of the laminate theory including the transverse shear effects, the transverse shear stresses in layer k of laminate is deduced [20, 21] from:
C45 yz yz C44 C xz k C45 55 k xz , where Cij are the transverse shear stiffness of layer k, and
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(212)
98
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
w
0 y C C yz 44 y 45 k C55 w0 xz k C45 x
Functions
x , y and w0
x .
(213)
are functions of coordinates (x, y). So, Expression (213)
shows that the transverse shear stresses are uniform through the layer thickness and discontinuous between, according to laminate theory. A better estimate can be obtained considering the governing equations of the mechanics of materials:
xxk xyk xzk 0, x y z yyk xyk yzk 0. y x z
(214)
These expressions allow us to derive the transverse shear stresses functions of in-plane stresses
xxk , yyk and xyk .
xzk and yzk
as
Considering Expression (199),
Equations (214) show that the transverse shear stresses are quadratic functions of the z coordinate. Moreover the transverse shear stresses are continuous at the layer interfaces and are zero on the two outer faces of the laminate. Finite element analysis gives the values of the transverse shear stresses ( yzlk , Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
xzlk , yzuk , xzuk ) on the lower and upper faces of each layer of each element e. So, the transverse shear stresses can be expressed as:
rk ark ( x, y ) z 2 brk ( x, y ),
r yz, xz,
(215)
where the coefficients are deduced from the values of the shear stresses on the lower and upper faces. Whence:
rk rk rk z 2 ,
r yz, xz,
(216)
with
rk
r lk r uk
hk hk 1 ek
,
rk ruk rk hk2 ,
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
r yz, xz. (217)
Damping in Composite Materials and Structures
99
7.6. Transverse Shear Strain Energy as Function of Transverse Shear Stresses The transverse shear strain energy for a given element e can be expressed in the material directions as: e e Use U44 U55 ,
(218)
with
1 2
e U 44 e U 55
1 2
4 4
d x d y d z,
5 5
d x d y d z,
e
e
(219)
where the integration is extended over the volume of the finite element e.
4 and 4
are
respectively the transverse shear stress and strain in plane ( T , T ) of material in layer k.
5 and 5 are the transverse shear stress and strain in plane ( L, T ) of material. The transverse shear strains and stresses are related by:
4 GTT 4 ,
(220)
are the transverse shear moduli in planes ( T , T ) and ( L, T ), respectively. It results that the transverse shear strain energies (219) can be written as: where
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
5 GLT 5 ,
GTT and GLT
1 2
e U 44
e U 55
In each layer k, stresses
e
42 d x d y d z, GTT
e
52 d x d y d z. GLT
1 2
4 and 5 ,
(221)
related to the material axes of the layer, can be
expressed as functions of the transverse shear stresses
yz and xz
in the finite element
directions (x, y, z) according to the stress transformations [20, 21]:
4 cos sin 5
sin yz cos xz .
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(222)
100
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. So, the transverse shear strain energies (221) are expressed as: e U 44
e U55
1 2
1 2
e
1 2 2 cos2 xz sin 2 2 xz yz sin cos dx dy dz, GTT yz (223)
e
1 2 2 sin 2 xz cos2 2xz yz sin cos dx dy dz, GTT yz (224)
Considering Expressions (223) and (224), the transverse shear energies can be expressed as: n
e U 44
n
e U 55
e U 44 k,
k 1
U55e k , k 1
(225)
e where Ursk ( rs 44, 55) are the transverse shear energies stored in layer k of
the
element e. Introducing the terms: e Uyzyzk
Se 2
hk
2 yzk d z,
hk 1
e Uyzxzk Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
e Uxzxz k
Se 2
hk
Se 2
hk
2 xzk d z,
hk 1
yzk xzk d z ,
hk 1
(226)
the transverse shear energies stored in layer k are expressed as:
e U 44 k
1 e e Ue cos2 k Uxzxzk sin 2 k 2Uyzxzk sin k cos k , GTT yzyzk (227)
e U55 k
1 e e Ue sin 2 k Uxzxzk cos 2 k 2Uyzxzk sin k cos k . GLT yzyzk (228)
7.7. Evaluation of Transverse Shear Strain Energy The energy terms
e e e Uyzyzk and Uxzxzk and Uyzxzk expressed in Equations (226) are
derived by introducing Expression (215) of the transverse shear stresses. It results that the energy terms can be written as:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures e U rsk
Se e I , 2 rsk
101
r , s yz, xz, (229)
with
1 1 e I rsk rk sk hk5 hk51 rk sk sk rk hk3 hk31 rk rk hk hk 1 , 5 3 r , s yz, xz. (230) Constants
rk , rk , sk and sk
are deduced from Equations (217).
Next, the transverse shear strain energies stored in element e are given by Expressions (225) and the total transverse shear strain energies stored in the finite element assemblage is obtained by summation on the elements as:
U 44
U55
e U 44 ,
elements
e U55 .
elements
(231)
7.8. Structural Damping and Discussion
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The damping of the finite element assemblage can be evaluated by extending the energy formulation approach considered in Section 5.2. The total strain energy stored in the laminated structure is given by:
Ud U11 U 22 2U12 U66 U 44 U55 ,
(232)
U11, U 22 , 2U12 and U66 are expressed by Equations (211), and the transverse shear strain energies U 44 and U 55 are given by Equations (231). where the in-plane strain energies
Then, the energy dissipated by damping in the layer k of the element e is derived from the strain energy stored in layer as: e e e e e e e e Uke 11 kU11k 22 kU 22 k 2 12 kU12 k 66 kU 66 k e e e e 44 kU 44 k 55 kU 55 .
(233)
introducing the specific damping coefficients pqk of the layer. This coefficients are related e
to the material directions ( L, T , T ) of layer: 11k and 22 k are the damping coefficients e
e
in traction-compression in the L direction and T direction of layer, respectively; 12 k is the
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
e
102
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
in-plane coupling coefficient; 66 k is the in-plane shear coefficient; 44 k and 55k are e
e
e
the damping coefficients in planes ( T , T ) and ( L, T ), respectively. The damping energy dissipated in the element e is next obtained by summation on the layers of element as: n
Ue
Uke , k 1
(234)
and the total energy U dissipated in the finite element assemblage is then obtained by summation on the elements:
U
U e .
elements
(235)
Finally, the damping of the finite element assemblage is characterised by the damping coefficient of the assemblage derived from relation:
U Ud .
(236) e
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
As observed in Section 5, the in-plane coupling energy U12 is much lower than the other in-plane energies and can be neglected.
7.9. Procedure and Discussion A general procedure was implemented to evaluate the damping of a structure using finite element analysis. This procedure is based on the previous formulation and can be applied to any structure for which the damping characteristics are different according to the layers and the elements of the assemblage. In the procedure, the finite element analysis is used first to establish [36] the eigen equations of the vibrations. The equation is solved to obtain the natural frequencies and the corresponding mode shapes. Next, the stresses, for each mode of vibration, on the lower and upper faces of each layer are read in each element of the finite element assemblage. The different energies are calculated according to the formulation developed in the previous sections and the damping
i
for each mode i is evaluated
according to Equation (236). The formulation considered is based on the laminate theory including the transverse shear effects. Results that we derived from the application of this formulation have shown that this formulation can be applied to all the composite materials considered by the authors: laminate materials, sandwich materials and laminate materials with interleaved viscoelastic layers.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
103
In the case of laminate materials, the results deduced from the formulation show that the transverse shear strain energies can be neglected with regard to in-plane strain energies. So, the damping is induced by the in-plane behaviour of laminate layers. In the case of sandwich materials, the results derived show that the behaviour of materials obtained by considering finite elements based on the laminate theory including the transverse shear effects is the same as the behaviour obtained by considering finite elements based on the sandwich theory [20, 21]. Moreover, the results obtained show that the transverse shear strain energies are much higher than the in-plane strain energies. Damping in sandwich materials is induced by the transverse shear behaviour of sandwich core. A similar behaviour as in the case of sandwich materials is observed in the case of laminate materials with interleaved viscoelastic layers. Damping is induced by the transverse shear behaviour of the viscoelastic layers. Finally, the results deduced from the previous finite element damping formulation is general and can be applied to laminate materials, sandwich materials and laminate materials with interleaved viscoelastic layers.
8. Experimental Investigation and Discussion on the Damping Properties
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
8.1. Materials The materials considered in the experimental investigation are laminate materials and sandwich materials. The laminate materials are constituted of E-glass fibres in an epoxy matrix and were fabricated with two different layers: unidirectional layers and serge weave layers. The weights of unidirectional fabrics and serge fabrics are 300 gm–2. Unidirectional Kevlar fibre laminates were also investigated by the authors [10, 18]. The experimental results obtained are similar with a damping somewhat greater in the case of Kevlar laminates. Laminate materials were prepared by hand lay-up process from epoxy resin with hardener and glass fabrics. Plates of different dimensions were cured at room temperature with pressure using vacuum moulding process, and then post-cured in an oven. The plates were fabricated with 8 layers in such a way to obtain the same plate thickness (nominal value of 2.4 mm) with the same reinforcement volume fraction (nominal value of 0.40). The engineering constants of laminates referred to the material directions ( L, T , T ) or (1, 2, 3) were measured in static tests as mean values of 10 tests for each constant. The values obtained are reported in Table 2. Then, the values of the reduced stiffnesses were derived and are reported in Table 3. Sandwich materials were constructed with [0/90]s cross-ply laminates as skins and with PVC closed-cell foams supplied in panels of thickness of 15 mm. Three foams were considered differing in their densities: 60 kg m–3, 80 kg m–3 and 200 kg m–3. The layers of the cross-ply laminates of the skins were constituted of the unidirectional layers considered previously. Mechanical characteristics of the foams were measured in static tensile tests for the Young’s modulus and the Poisson’s ratio, and in static shear tests for the shear modulus. The values derived are reported in Table 4.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
104
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. Table 2. Engineering constants of laminates Laminate Unidirectional layer Serge layer
EL (GPa) 29.9 16
νLT 0.24 0.24
ET (GPa) 7.50 15.4
GLT (GPa) 2.25 2.10
Table 3. Reduced stiffnesses of laminates Laminate Unidirectional layer Serge layer
Q11 (GPa) 30.3 16.9
Q12 (GPa) 1.83 3.91
Q22 (GPa) 7.61 16.3
Q66 (GPa) 2.25 2.10
Q16 0, Q26 0. Table 4. Mechanical characteristics of the foams Density (kg m–3) 60 80 200
Young’s modulus (MPa) 59 83 240
Poisson’s ratio 0.42 0.43 0.45
Shear modulus (MPa) 22 30 80
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
8.2. Experimental Equipment The equipment used is shown in Figure 6. The test specimen is supported horizontally as a cantilever beam in a clamping block. An impulse hammer is used to induce the excitation of the flexural vibrations of the beam. A force transducer positioned on the hammer allows us to obtain the excitation signal as a function of the time. The width of the impulse and hence the frequency domain is controlled by the stiffness of the head of the hammer. The beam response is detected by using a laser vibrometer which measures the velocity of the transverse displacement of a point near the free end of the beam. Next, the excitation and the response signals are digitalized and processed by a dynamic analyzer of signals. This analyzer associated with a PC computer performs the acquisition of signals, controls the acquisition conditions (sensibility, frequency range, trigger conditions, etc.), and next performs the analysis of the signals acquired (Fourier transform, frequency response, mode shapes, etc.). Then, the signals and the associated processings can be saved for post-processings. The system allows the simultaneous acquisition of two signals with a maximum sampling frequency of 50 kHz with a resolution of 13 bits for each channel.
8.3. Analysis of the Experimental Results 8.3.1. Determination of the Constitutive Damping Parameters Impulse excitation of the flexural vibrations of beam was induced (Figure 7) at point x1 near the clamping block and the beam response was detected at point x near the free end of the beam. Figure 8 gives an example of the Fourier transform of the beam response to an
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
105
impulse input. This response shows peaks which correspond to the natural frequencies of the bending vibrations of the beam. Experimental analysis was performed on beams of different lengths 160, 180 and 200 mm so as to have a variation of the values of the peak frequencies.
Figure 6. Experimental equipment for damping analysis.
l
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
x x x1 Measuring point
Impact point
Figure 7. Impact and measuring points on the cantilever beams.
The transverse response to an impact loading is given by expression (58). In fact, the laser vibrometer measures the velocity of the transverse displacement and the beam response detected by the vibrometer is proportional to:
w w 0 0 p1 t
X i( x1) X i( x) i2 Ki( ) (ai sin t bi cos t ). i 1
(237)
The Fourier transform gives the complex amplitude as function of the frequency expressed by:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
106
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
W0 ( ) p1
X i( x1) X i( x) i2
ai2 bi2 Ki ( ).
i 1
(238)
70 60
Amplitude (dB)
50 40 30 20 10 0 0
100
200
300
400
500
600
700
Frequency (Hz)
Figure 8. Typical frequency response to an impulse of a unidirectional glass composite beam.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
So, the experimental analysis was implemented by fitting the experimental responses with relation (238), considering either the viscous friction model (Equations (37) and (38)) or the complex stiffness model (Equations (49) and (50)). This fitting was obtained by a least square method using the optimisation toolbox of Matlab, which allows us to derive the values of the natural frequencies fi, and the modal damping coefficient ξi (case of damping using viscous friction modelling) or the loss factor ηi (case of damping using the complex stiffness model). This method can be applied for notable damping of materials. According to Relations (21), each natural frequency of the undamped beam is related to the stiffness by unit area by the relation:
4 2 fi 2
i4 k s . a4 s
(239)
This relation allows us to evaluate the stiffness ks for each natural frequency of beams in the case of low damping. Expressions (237) and (238) were established in the case of orthotropic laminates. In the case of beams made of sandwich materials, the fitting of the experimental responses were implemented by two procedures. The first procedure used the frequency analysis of Matlab Toolbox. The second one fits the experimental responses with the responses obtained by finite element analysis.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
107
8.3.2. Plate Damping Measurement Rectangular plates with two adjacent edges clamped with the other two free and plates with one edge clamped and the others free were also tested to determine the damping characteristics for the first modes of flexural vibrations. As in the case of beams, the excitation of vibrations was induced by the impulse hammer and the plate response was detected by using the laser vibrometer. The damping parameters were derived from the Fourier transform of the plate response. Vibration excitation and response detection were carried out at different points of the plates so as to generate and detect all the modes.
8.4. Damping of Unidirectional Laminates 8.4.1. Experimental Results As reported previously, the experimental investigation of damping was performed on beams of different lengths: 160, 180 and 200 mm so as to have a variation of the values of the peak frequencies. Beams had a nominal width of 20 mm and a nominal thickness of 2.4 mm Fitting the experimental responses of beams with the analytical responses (Subsection
= 60 ° = 45 ° = 75 ° = 90 ° = 30 °
1.7 1.5
Loss factor i (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
8.3.1) leads to the evaluation of the modal damping coefficient ξi or the loss factor ηi, associated to each mode i. Figure 9 shows the experimental results obtained in the case of glass fibre composites for the loss factor. The results are reported for the first three bending modes and for the different lengths of the beams. The experimental results show that damping is maximum at a fibre orientation of about 60°.
1.3 1.1
= 15 °
0.9 0.7
=0° 0.5 0.3 0
200
400
600
800
Frequency (Hz) Figure 9. Experimental results obtained for the damping as a function of the frequency for different fibre orientations, in the case of unidirectional glass fibre composites.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
108
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. Table 5. Damping increase (%) in the frequency range [50, 600 Hz] Fibre orientation (°) Glass fibre composites
0 21
15 24
30 26
45 23
60 26
75 23
90 27
For a given fibre orientation, it is observed that the damping increases when the frequency is increased. The values of the damping increase when the frequency is increased from 50 Hz to 600 Hz are reported in Table 5 for the unidirectional glass fibre composites. The table shows that the damping increase is fairly the same for the different fibre orientations: from 21 to 27 %.
8.4.2. Comparison of Experimental Results and Models 8.4.2.1. Models of Adams-Bacon and Ni-Adams The models are based on an energy analysis and lead to the evaluation (99) of the specific damping coefficient ψ measured in the direction θ as function of the damping coefficients ψ11 in the 0° direction, ψ22 in the 90° direction and ψ66 the damping coefficient associated to inplane shear. It is usual to consider the results obtained for the loss factor η related to ψ by the relation ψ = 2πη. Thus, the formulation (99) is simply transposed by considering the loss factors η11, η12 and η66. The values of these coefficients can be derived from the experimental results by considering the results obtained for fibre orientations of 0° and 90°, and for an intermediate orientation of 45°, for example. The analytical curve giving the damping η (θ) as a function of the fibre orientation is then derived using Equation (99).
1.4 Loss factor i (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
1.6
1.2 1.0 0.8 0.6
Experimental results Adams-Bacon analysis Ni-Adams analysis Complex stiffness model
0.4 0.2 0
10
20
30 40 50 60 Fibre orientation (°)
70
80
90
Figure 10. Comparison between the experimental damping results and the results derived from AdamsBacon, Ni-Adams and complex stiffness models, in the case of glass fibre composites.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
109
The results deduced from the Adams-Bacon and Ni-Adams models are compared with the experimental results at frequency 50 Hz in Figure 10. The curve derived from the AdamsBacon model is obtained with:
11 0.40%,
22 1.24%,
66 1.48%.
(240)
The one deduced from the Ni-Adams model is obtained with:
11 0.40%,
22 1.24%,
66 1.72%.
(241)
In Figure 10, it is observed a rather good agreement between the results deduced from the two models and the experimental results. However, the values of the shear loss factor deduced from the two models are fairly different. 8.4.2.2. Complex Stiffness Model The damping evaluation using the complex modulus of the beams was considered in Subsection 4.2. The damping is evaluated by relation (103), where the complex bending * 1 modulus is expressed (102) as function of the element D11 of the complex inverse matrix of Dij* . According to the elastic-viscoelastic correspondence principle, the complex
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
bending-twisting matrix Dij* is obtained as: 3 Dij* h Qij * , 12
where the complex reduced stiffnesses the reduced stiffnesses stiffnesses
(242)
Qij * are converted from the relations of Table 1 giving
Qij with reference to the fibre orientation as functions of the reduced
Qij referred to the material directions. Thus, the complex reduced stiffnesses in
the material directions are expressed as: * Q11
EL*
,
* * 2 ET 1 LT *
EL
* Q22
ET* *
2 ET *
* 1 LT
* Q12
* * TL EL * * 2 ET 1 LT *
,
EL
,
* * Q66 GLT ,
EL
introducing the engineering moduli in the complex form: Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(243)
110
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
EL* EL 1 i L ,
ET* ET 1 i T ,
* LT LT 1 i LT .
* GLT GLT 1 i LT ,
(244)
When the fibre orientation is equal to 0°, the effective bending modulus can be identified with the longitudinal modulus EL of the material. Hence, the longitudinal loss factor can be identified with the damping η0° measured for the 0° fibre orientation. In the same way, the transverse loss factor can be identified with the loss factor η90° measured for the 90° fibre orientation. The results obtained by the complex stiffness model are reported in Figure 10 in the case of the unidirectional glass fibre composites. The results were obtained by considering that the damping associated to the Poisson's ratio is zero and fitting the shear loss factor ηLT so that the complex stiffness model gives the value of the loss factor measured for the 45° fibre orientation. The comparison between the results obtained shows that the experimental results are not well described by the complex stiffness model for fibre orientations ranging from about 10° to 45°. 8.4.2.3. Using the Ritz Method
8.4.2.3.1. Damping Parameters
2.0 1.8
Loss factor i (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The analysis using the Ritz method (Section 5) was applied to the experimental results obtained for the bending of beams. The beams were considered in the form of plates with one edge clamped and with the others free. Damping was evaluated by the Ritz method (140) considering the beam functions (Subsection 5.1.2). Thus, the present evaluation of the beam
1.6 1.4 1.2 f = 50 Hz f = 300 Hz f = 600 Hz f = 50 Hz f = 300 Hz f = 600 Hz
1.0 Experimental results 0.8 0.6
Ritz analysis
0.4 0
10
20
30
40
50
60
70
80
90
Fibre orientation (°)
Figure 11. Comparison of the experimental results and the results deduced from the Ritz method for damping as function of fibre orientation, in the case of glass fibre composites.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
111
damping takes account of the effect of the beam width. The results deduced from the Ritz method are reported in Figures 11 in the case of unidirectional glass fibre composites. A good agreement is obtained with the experimental results. The values of the loss factors considered for modelling are reported in Table 6 for the frequencies 50, 300 and 600 Hz. These results show that the shear damping evaluated by using the Ritz method is fairly higher that the values of the shear loss factor deduced from the Adams-Bacon analysis or from the Ni-Adams analyses which do not consider the width of the beam. Table 6. Loss factors derived from the Ritz method in the case of unidirectional glass fibre composites f (Hz) 50 300 600
η11 (%) 0.35 0.40 0.45
η12 0 0 0
η22 (%) 1.30 1.50 1.65
η66 (%) 1.80 2.00 2.22
1.6
Loss factor i (%)
1.4 1.2 1.0 0.8
Experimental results R = 100 R = 20 R = 10 R=7 R=5
0.6
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
0.4 0.2 0
10
20
30
40
50
60
70
80
90
Fibre orientation (°)
Figure 12. Unidirectional beam damping obtained with different values of the length-to-width ratio R of the beam, in the case of glass fibre composites.
8.4.2.3.2. Influence of the Width of the Beams The influence of the beam width can be analyzed by the Ritz method. Figure 12 shows the results obtained for the loss factor of the first mode of beams with a nominal length of 200 mm and for different length-to-width ratio of the beam: 100, 20, 10, 7 and 5, in the case of unidirectional glass fibre composites. The figure shows that the results reach a limit for high values of the length-to-width ratio of the beams. Furthermore, comparison of the results deduced from the Ni-Adams analysis with the results derived from the Ritz analysis shows that the Ni-Adams analysis can be applied to the evaluation of damping properties of beams with high values of the length-to-width ratio. In fact, in order to minimize the edge effects especially for off-axis materials it is difficult to implement an experimental analysis with a high value of the length-to-width ratio of the beams. A ratio about 10 which leads to a beam
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
112
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
width of 20 mm for a length of 200 mm appears to be a good compromise. In this case it is necessary to analyse the experimental results with a modelling which takes the width of the beams into account.
8.4.2.3.3. Damping according to the modes of beam vibrations The Ni-Adams analysis is established using the beam theory which considers the case of bending along the x axis of beams and assumes that the transverse displacement of beams is a function of the x coordinate only w0 = w0(x).
(245)
According to this theory, only the bending modes of beams are described and the damping of unidirectional beams will all the more high as the beam deformation will induce bending in the direction transverse to fibres and in-plane shearing for intermediate orientations of fibres. The Ni-Adams analysis does not take account of the effects of beam twisting which can induce notable twisting deformation of beams for which the transverse displacement is not anymore independent of the y coordinate. 2.2 2.0
Loss factor i (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
1.8 1.6 1.4 1.2 1.0
mode 1 mode 2 mode 3 mode 4 mode 1 mode 2 mode 3
Ritz analysis
0.8 0.6
Experimental bending modes
0.4 0.2 0
10
20
30
40
50
60
70
80
90
Fibre orientation (°)
Figure 13. Variation of the damping of unidirectional beams of length equal to 180 mm, derived from the Ritz method for the first four modes of unidirectional glass fibre beams.
Figure 13 shows the variations of beam damping deduced from the Ritz method for the first four modes of unidirectional beams in the case of beam length equal to 180 mm and a length-to-width ratio equal to 10, for unidirectional glass fibre beams. For the damping evaluation of beams we have considered that the loss factors of the materials depend on the frequency according to the results obtained in Subsection 8.4.2.3.1. The natural frequencies and modes of the beams were first derived using the Ritz method. Next, the damping evaluation of laminated beams was derived according to the modelling developed in Section 5 and considering that the damping loss factors η11, η22 and η66 increased linearly in the
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
113
frequency range [50, 600 Hz] according to the values reported in Table 6. The results for the first two modes are similar, differing by the increase of the damping with the frequency.
mod e1
mod e2
mod mod e3 e4 Figure 14. Free flexural modes of a unidirectional glass fibre beam for 0° fibre orientation.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
mod e1
mod e2
mod mod e3 e4 Figure 15. Free flexural modes of a unidirectional glass fibre beam for 30° fibre orientation.
In the case of the third mode (Figure 13), it is observed a high beam damping for fibre orientations of 0° and 10° with a value which is fairly near of the shear damping. The shapes of the modes 1 to 4 for a fibre orientation of 0° are given in Figure 14. The results show that the shapes of modes 1, 2 and 4 satisfy the assumption (245), whereas an important twisting of the beam is observed for the mode 3 inducing a notable in-plane shear deformation. Finally, the beam damping results from the respective contributions of the energies induced in bending along the x direction of the beam, bending along the transverse y direction and beam twisting. These energies are taken into account by the damping analysis based on the Ritz
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
114
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
method. Figure 15 reports the mode shapes deduced in the case of 30° fibre orientation showing the participation of the different deformation modes. In this case, it is observed that beam twisting of the mode 4 is associated to a lower damping of the beam. These results show that beam twisting induces an increase of damping for fibre orientations near the material directions: 0° direction for mode 3 and 90° direction for mode 4 (Figures 13 and 14), resulting from the increase of in-plane shear deformation of materials. In contrast, the beam twisting results in a decrease of damping for intermediate orientations (mode 4, Figures 13 and 15) associated to the decrease of in-plane shear deformation. The variations of beam damping deduced from the Ritz method are compared in Figure 14 with the experimental results obtained for the first three bending modes of beams. These bending modes were obtained by exciting the beams by an impulse applied on the beam axis so as to induce vibration modes without beam twisting. The experimental results agree fairly well with the damping evaluation by the Ritz method when only the bending modes of the beams are considered.
8.5. Damping of Laminated Beams Laminated beams with three different stacking sequences were analyzed: [0/90/0/90]s
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
cross-ply laminates, [0/90/45/−45]s laminates and [θ/−θ/θ/−θ]s angle-ply laminates with θ varying from 0° to 90°. The laminates were prepared from 8 plies of the unidirectional materials studied in the previous section. The nominal thickness of the laminates was 2.4 mm and the analysis was implemented in the case of beams 200 mm long and 20 mm width. Figure 16 shows the results obtained for the damping in the case of glass fibre laminates. Figure reports the results deduced for the damping by the Ritz method for the first four modes and the experimental damping measured for the first mode. The evaluation of laminate damping by the Ritz method takes account of the variation of the loss factors η11, η22 and η66 with frequency (Table 6). For the cross-ply laminates (Figure 16a) and [0/90/45/−45]s laminates (Figure 16b), the material damping is derived as a function of laminate orientation. For the [θ/−θ/θ/−θ]s angle ply laminates (Figures 16c), damping is reported as a function of the ply orientation θ. The damping deduced from the Ritz method was evaluated by applying the results of Section 5.2 to the different laminates. The in-plane behaviour of the [0/90/0/90]s cross-ply laminates is the same in the 0° and 90° directions, when the external 0° layers of the stacking sequence leads to a slight increase of the bending properties in the 0° direction. Thus, compared to the damping of unidirectional composites (Figure 13), the stacking sequence [0/90/0/90]s leads to a more symmetric variation of damping (Figure 16a) as a function of the orientation with damping characteristics which are slightly higher in the 90° direction. Near 45° orientations damping of the [0/90/0/90]s laminates is clearly reduced (about 1.2 % for the first two modes) compared to the damping of the unidirectional laminates (about 1.4 %). This reduction results from the in-plane shear deformation which is constrained by the [0/90] stacking sequence. For the third mode it is observed a high damping for directions near 0° and 90° associated to the effects of beam twisting as in the case of the unidirectional laminates. For the fourth mode the beam twisting
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
115
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
leads to a decrease of the beam damping. The use of the [90/0/90/0]s stacking sequence would lead to a damping behaviour where the 0° and 90° directions would be inverted.
Figure 16. Damping variation as function of laminate orientation for beams of different glass fibre laminates: a) [0/90/0/90]s cross-ply laminates, b) [0/90/45/−45]s laminates and c) [θ/−θ/θ/−θ]s angle ply laminates. Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
116
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
For [0/90/45/−45]s laminates (Figure 16b), the damping behaviour is practically symmetric as a function of the fibre orientation with an in-plane shear constrain effect which is more important than in the case of cross-ply laminates, leading to a reduction of the damping near 45° orientation, for modes 1 and 2: loss factor of about 0.98 % in the case of mode 1. In the case of the [θ/−θ/θ/−θ]s angle ply laminates and for the first three modes (Figure 16c), the damping for ply angles higher than 60° is practically the same as damping observed for the unidirectional beams with fibre orientation equal to θ. For lower values of ply angle, it is observed a reduction of laminate damping comparatively to the unidirectional composites, associated to the in-plane constrain effect induced by the [θ/−θ/θ/−θ]s sequence. For mode 4 the damping reduction of angle ply laminates is observed for all the ply orientations, except for orientations near 0° and 90° where angle ply laminates are similar to unidirec-tional laminates.
8.6. Damping of Cloth Reinforced Laminates Figure 17 shows the experimental results obtained for the damping in the case of glass serge composites. The results are reported for the first three bending modes and for the different lengths of the beams as functions of the frequency and for different orientations of glass fibres. For a given serge orientation, it is observed that damping increases when the frequency is increased. The values of the damping increase when the frequency is increased from 50 Hz to 600 Hz are reported in Table 7. The increase is fairly higher (Table 5) in the case of unidirectional laminates (21 to 26 %) than in the case of serge laminates (17 to 21 %). 1.6
= 45 = 30= 60
Loss factor i (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
1.4 1.2
= 15= 75 1.0
= 0= 90
0.8 0.6 0.4 0
200
400
600
800
1000
1200
Frequency (Hz)
Figure 17. Experimental results obtained for the damping as a function of the frequency for different orientations, in the case of glass serge composites.
Table 7. Damping increase (%) in the frequency range [50, 600 Hz] Fibre orientation (°) Glass serge composites
0 18
15 19
30 21
45 17
60 21
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
75 19
90 18
Damping in Composite Materials and Structures
117
The variations of the loss factor with serge orientation are given in Figure 18 for the three frequencies 50, 300 and 600 Hz. In the case of the unidirectional glass fibre laminates, the transverse damping is higher than the longitudinal damping, and the damping is maximum (Figure 11) at a fibre orientation of about 60° for the glass fibre composites. In the case of serge laminates, the damping variation is symmetric with a damping maximum for the orientation of 45°. As in the case of unidirectional laminates, the analysis using the Ritz method (Section 5) or the analysis using the finite element method (Section 7) can be applied to the experimental results obtained for the bending of serge beams. The beams were considered in the form of plates with one edge clamped and with the others free. The results obtained are identical using either the Ritz analysis or the finite element analysis, and these results are reported in Figure 18. A good agreement is obtained with the experimental results. The values of the loss factors considered for modelling are reported in Table 8 for the frequencies 50, 300 and 600 Hz. The increase of fibre number in the 90° direction from the unidirectional laminates to the serge laminates leads to an increase of the loss factor 11 in the 0° direction, a decrease of the loss factor 22 in the 90° direction and a decrease of the shear loss factor 66 . 1.6
Loss factor i (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
1.4 1.2 1.0 0.8
f = 50 Hz f = 300 Hz f = 600 Hz f = 50 Hz f = 300 Hz f = 600 Hz
Experimental results
0.6
Modelling
0.4 0
10
20
30
40
50
60
70
80
90
Fibre orientation (°)
Figure 18. Variation of the loss factor as function of orientation in the case of glass serge composites. Comparison between experimental results and modelling.
Table 8. Loss factors derived from modelling in the case of glass serge laminates f (Hz) 50 300 600
η11 (%) 0.67 0.83 0.89
η12 0 0 0
η22 (%) 0.67 0.83 0.89
η66 (%) 1.53 1.78 1.83
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
118
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. 1.0
Loss factor i (%)
0.9 0.8 0.7 0.6 Serge [(0/90)2]s
0.5
[02/902]s 0.4 0
200
400
600
800
1000
1200
Frequency (Hz)
Figure 19. Comparison between damping of serge laminates and damping of cross-ply laminates, for 0° orientation of the laminates.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 19 compares the results obtained for damping in the case of serge laminates and cross-ply laminates, for 0° orientation of the laminates. Two cross-ply laminates are considered: [(0/90)2]s and [02/902]s. Damping of [(0/90)2]s laminates is slightly higher than that of [02/902]s laminates. This fact results from the damping of the 90° layers which are more external in the [(0/90)2]s laminates. In Figure 19, it is observed that the damping of serge laminates is clearly higher than the damping of cross-ply laminates. This increase of damping may be associated with the energy which is dissipated by friction between the warp fibres and weft fibres, in the case of the serge laminates.
8.7. Damping of Unidirectional Laminates with Interleaved Viscoelastic Layers 8.7.1. Materials The materials investigated are the unidirectional glass fibre composites, considered in Section 8.4, in which a single or two viscoelatic layers are interleaved. The volume fraction of fibres is equal to 0.40 and the nominal thickness e of the unidirectional layers is 2.4 mm. The engineering constants and the modal loss factors η (related to the specific damping coefficients ψ by relation: 2 ) referred to the material directions were evaluated in Sections 8.1 and 8.2. The values are reported in Table 7. The viscoelastic layers are constituted of Neoprene based layers of nominal thickness e0 = 0.2 mm. Three types of laminates have been investigated: a laminate with a single viscoelastic layer of thickness e0 interleaved in the middle plane, a laminate with a single viscoelastic layer of thickness 2e0 in the middle plane and a laminate with two viscoelastic layers of thickness e0 interleaved at the distance e/2 from the middle plane. Plates were hand laid up and cured at room temperature with a pressure of 70 kPa using vacuum moulding process.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
119
In the case of a single interleaved viscoelastic layer, the nominal thickness of the laminates is 2.6 mm with a weight of 3.7 kgm–2. Interleaving two viscoelastic layers leads to a laminate thickness of 2.8 mm with a weight of 3.85 kgm–2. Aluminium spacers of the same thicknesses as the viscoelastic layers were added in the root section between the unidirectional layers. Beam specimens were next cut from the plates and the damping properties were measured for different orientations of the fibres.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
8.7.2. Experimental Results The damping of the materials was deduced from impulse tests using the experi-mental procedure described in Sections 8.2 and 8.3. The experimental evaluation of damping was performed on beams 20 mm wide of different lengths: 160, 180 and 200 mm, so as to have a variation of the values of the natural frequencies of the beams. Only the first two modes were considered. Figures 20, 21 and 22 report the experimental results obtained for the beam damping as a function of the fibre orientation for the three beam lengths (Figures 20a, 21a and 22a) and the beam damping as a function of the frequency for the different fibre orientations (Figures 20b, 21b and 22b). The results are given in the case of laminates with a single viscoelastic layer of thickness e0 = 0.2 mm (Figure 20), laminates with a single viscoelastic layer of thickness e0 = 0.4 mm (Figure 21) and laminates with two viscoelastic layers of thicknesses e0 = 0.2 mm (Figure 22). The experimental results obtained for the unidirectional materials without viscoelastic layers are also reported (Figures 20a, 21a and 22a). The experimental results show that the damping of laminates increases significantly upon interleaving a single or two viscoelastic layers. The fibre orientation dependence of damping appears somewhat similar to that of the laminates without viscoelastic layers, but with a damping maximum which is moved from 60° fibre orientation to 30° fibre orientation when viscoelastic layers are interleaved. Moreover in contrast to the non-interleaved laminates, damping increases significantly with frequency depending on the vibration mode. In the case of a single viscoelastic layer interleaved in the middle plane, the laminate damping is increased all the more since the viscoelastic layer is thick. The damping of laminate with two interleaved viscoelastic layers of thicknesses e0 is lower (about 1.6 time) than the one measured in the case of laminate with a single viscoelastic layer of thickness 2e0, when the damping is fairly similar to the damping of laminate with a single layer of thickness e0. This results from the fact that the energy is essentially dissipated by transverse shear of the viscoelastic layers and the associated energy is maximum in the middle plane of laminates. Table 9. Properties of the glass fibre composites without viscoelastic layers EL (GPa) 29.9
ET (GPa) 5.85
GLT (GPa) 2.45
νLT 0.24
η11, ηL (%) 0.40
η22, ηT (%) 1.50
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
η66, ηLT (%) 2.00
120
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. 6 1st mode 2nd mode 1st mode 2nd mode 1st mode 2nd mode without viscoelastic layer
l = 160 mm
5
Loss factor (%)
l = 180 mm l = 200 mm
4 3 2 1 0 0
10
20
40
50
60
1st mode
Loss factor (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
90
3
2
1
80
2nd mode
5
4
70
Fibre orientation (°)
(a)
6
30
0
(b)
100
200
300
400
500
600
Frequency (Hz)
Figure 20. Experimental results obtained in the case of glass fibre composites with a single viscoelastic layer of thickness 0.2 mm interleaved in the middle plane and for three lengths of the test specimens: a) laminate damping as function of the fibre orientation and b) laminate damping as function of the frequency.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
121
10 1st mode 2nd mode 1st mode 2nd mode 1st mode 2nd mode without viscoelastic layer
l = 160 mm
Loss factor (%)
8
l = 180 mm l = 200 mm
6
4
2
0 0
10
20
30
40
50
60
70
80
90
Fibre orientation (°)
(a)
9 1st mode
2nd mode
8
Loss factor (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
7
6
5
4
3
2
1 0
(b)
100
200
300
400
500
600
Frequency (Hz)
Figure 21. Experimental results obtained in the case of glass fibre composites with a single viscoelastic layer of thickness 0.4 mm interleaved in the middle plane and for three lengths of the test specimens: a) laminate damping as function of the fibre orientation and b) laminate damping as function of the frequency.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
122
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. 6 1st mode 2nd mode 1st mode 2nd mode 1st mode 2nd mode without viscoelastic layer
l = 160 mm
5
Loss factor (%)
l = 180 mm l = 200 mm
4 3 2 1 0 0
10
20
30
40
50
60
70
80
90
Fibre orientation (°)
(a)
6 1st mode
2nd mode
Loss factor (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
5
4
3
2 1
0 0
(b)
100
200
300
400
500
600
Frequency (Hz)
Figure 22. Experimental results obtained in the case of glass fibre composites with two viscoelastic layer of thickness 0.2 mm interleaved away from the middle plane and for three lengths of the test specimens: a) laminate damping as function of the fibre orientation and b) laminate damping as function of the frequency.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
123
8.7.3. Analysis of the Experimental Results
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
8.7.3.1. Dynamic Properties of the Viscoelastic Layers In the case of laminates with interleaved viscoelastic layers, the laminate damping can be evaluated either by the modelling using the Ritz method developed in Section 6.3 or by the finite element analysis considered in Section 7. This evaluation needs to obtain the values of the Young’s modulus and the loss factors of the viscoelastic layers. These characteristics depend on the frequency and are generally derived according to the standard ASTM E 756 [48]. Following this standard, the damping characteristics of the viscoelastic material were evaluated from the flexural vibrations of a clamped-free beam 10 mm wide and constituted of two aluminium beams with a layer of the viscoelastic material interleaved between the aluminium beams. An aluminium spacer was added in the root section between the two aluminium beams of the test specimens. The roots were machined as part of the aluminium beams to obtain a root section 40 mm long and 10 mm high and then the root section was closely clamped in a rigid fixture. The free length and the thicknesses of the aluminium beams were selected so as to measure the damping characteristics on the frequency range [50, 600 Hz] considered in the case of the experimental analysis of interleaved laminates (Subsection 8.7.2). Thus, the beam dimensions used were a free length varying from 200 to 300 mm, a thickness of the viscoelastic layer of 0.2 mm and thicknesses of aluminium beams of 1 mm. The Young’s modulus of the visco-elastic layer was deduced from the natural frequencies of the test specimens and the loss factor was evaluated by applying the results of the modelling considered in Section 7 to the case of the aluminium-viscoelastic layer laminates. Figures 23 and 24 report the experimental results obtained, using logarithmic scales for the Young’s modulus and for the frequency. In the frequency range studied, it is observed linear variations for the logarithm of the Young’s modulus and for the loss factor of the viscoelastic material. The results of Figures 23 and 24 lead to:
log Ev 0.106log f 1.52,
Ev (MPa),
(246)
for the variation of the Young’s modulus of the viscoelastic layer with the frequency, and:
v 39.4 5.56log f , v (%) , for the loss factor of the viscoelastic material.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(247)
124
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. 80
Young's modulus (MPa)
70
60
50
40 35 10
50
100
500
1000
Frequency (Hz) Figure 23. Frequency dependence of the Young’s modulus of the viscoelastic layers.
34
Loss factor
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
(%)
32 30 28 26 24 22 20 10
100
1000
Frequency (Hz) Figure 24. Frequency dependence of the loss factor of the viscoelastic layers.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
125
6 1st mode modelling 2nd mode 1st mode experiment 2nd mode without viscoelastic layer
Loss factor (%)
5 4 3 2 1 0 0
10
20
(a)
30
40
50
60
70
80
90
Fibre orientation (°) 5 1st mode modelling 2nd mode 1st mode experiment 2nd mode without viscoelastic layer
Loss factor (%)
4
3
2
1
0 0
10
20
30
40
50
60
70
80
90
Fibre orientation (°) 5 1st mode modelling 2nd mode 1st mode experiment 2nd mode without viscoelastic layer
4
Loss factor (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
(b)
3
2
1
0 0
(c)
10
20
30
40
50
60
70
80
90
Fibre orientation (°)
Figure 25. Comparison between the experimental results and the results deduced from the modelling, in the case of a single viscoelastic layer 0.2 mm thick, for test specimen lengths of : a) l = 160 mm, b) l = 180 mm, c) l = 200 mm.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
126
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. 10 1st mode modelling 2nd mode 1st mode experiment 2nd mode without viscoelastic layer
Loss factor (%)
8
6
4
2
0 0
10
20
(a)
30
40
50
60
70
80
90
Fibre orientation (°) 8 1st mode modelling 2nd mode 1st mode experiment 2nd mode without viscoelastic layer
7
Loss factor (%)
6 5 4 3 2 1 0 0
10
20
30
40
50
60
70
80
90
Fibre orientation (°) 7 1st mode modelling 2nd mode 1st mode experiment 2nd mode without viscoelastic layer
6
Loss factor (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
(b)
5 4 3 2 1 0 0
(c)
10
20
30
40
50
60
70
80
90
Fibre orientation (°)
Figure 26. Comparison between the experimental results and the results deduced from the modelling, in the case of a single viscoelastic layer 0.4 mm thick, for test specimen lengths of : a) l = 160 mm, b) l = 180 mm, c) l = 200 mm.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
127
6 1st mode modelling 2nd mode 1st mode experiment 2nd mode without viscoelastic layer
Loss factor (%)
5 4 3 2 1 0 0
10
20
(a)
30
40
50
60
70
80
90
Fibre orientation (°) 5 1st mode modelling 2nd mode 1st mode experiment 2nd mode without viscoelastic layer
Loss factor (%)
4
3
2
1
0 0
10
20
30
40
50
60
70
80
90
Fibre orientation (°) 5 1st mode modelling 2nd mode 1st mode experiment 2nd mode without viscoelastic layer
4
Loss factor (%)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
(b)
3
2
1
0 0
(c)
10
20
30
40
50
60
70
80
90
Fibre orientation (°)
Figure 27. Comparison between the experimental results and the results deduced from the modelling, in the case of two viscoelastic layers 0.2 mm thick, for test specimen lengths of : a) l = 160 mm, b) l = 180 mm, c) l = 200 mm.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
128
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
8.7.3.2. Damping of the Glass Fibre Laminates with Interleaved Viscoelastic Layers The loss factor of the glass fibre laminates with interleaved viscoelastic layers was derived from the modelling using the Ritz method (Section 6.3) and the damping modelling using finite element analysis (Section 7). The results obtained are very similar. The results derived from the finite element analysis are compared with the experimental results in Figures 25, 26 and 27, for the first two modes of the test specimens: in the case of a single interleaved viscoelastic layer 0.2 mm thick, for the different free lengths of the test specimens l = 160 mm (Figure 25a), l = 180 mm (Figure 25b) and l = 200 mm (Figure 25c); in the case of a single interleaved viscoelastic layer 0.4 mm thick, for the different free lengths of the test specimens l = 160 mm (Figure 26a), l = 180 mm (Figure 26b) and l = 200 mm (Figure 26c); in the case of two interleaved viscoelastic layers 0.2 mm thick, for the different free lengths of the test specimens l = 160 mm (Figure 27a), l = 180 mm (Figure 27b) and l = 200 mm (Figure 27c). It is observed that the results deduced from the modelling describe fairly well the experimental damping variation obtained as function of the fibre orientation. Furthermore the modelling results corroborate that damping of laminates with two viscoelastic layers of thicknesses e0 introduced at the quarters of the thickness of laminates (Figure 27) is equal to the damping of laminates with a single viscoelastic layer of thickness e0 interleaved in the middle plane (Figure 25).
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
9. Dynamic Response of a Damped Composite Structure As an application, modelling developed in Section 7 and the results deduced from the investigation developed in Section 8 for the damping of the different materials were applied [49] to the analysis of the simple shape structure of Figure 28. Three types of materials were used for the structure: glass serge laminate of thickness of 5 mm; glass serge laminate with interleaved viscoelastic layer 0.2 mm thick; sandwich material with PVC foam 15 mm thick and density of 60 kg m–3, and glass [0/90]s skins 1.2 mm thick. The damping properties of these different materials were analysed in the previous section. The different materials of the structure were chosen in such a way to have the same stiffness of the structure. The structure was clamped in a clamping block of dimensions 150 mm × 150 mm. An impulse hammer was used to induce the excitation of the vibrations of the structure. The response of the structure was detected by using a laser vibrometer. Different impact points and measuring points were considered to induce and to detect all the vibration modes of the structure. Figure 29 shows the shapes of the first six modes deduced from finite element analysis in the case where the structure is constituted of serge laminate or serge laminate with interleaved viscoelastic layer. In the case of sandwich material, modes 1 and 2 are inverted. Mode 1 is a twisting mode, mode 2 a longitudinal bending mode and mode 3 a transverse bending mode. The other modes combine these different effects.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
129
Clamping block
150 mm
560 mm
Impact point
Measuring point
80 mm 150 mm
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 28. Simple shape structure.
mode 1
mode 2
mode 3
mode 4
mode 5
mode 6
Figure 29. Examples of the shapes of the vibration modes of the structure constituted of serge laminate or serge laminate with interleaved viscoelastic layer. Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
130
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. Table 10. Comparison of the modal loss factors deduced from modelling and the ones obtained by experimental investigation, in the case of the structure constituted of the serge laminate
mode 1 mode 2 mode 3 mode 4 mode 5 mode 6 mode 7 mode 8 mode 9 mode 10
Modelling Mode frequency Loss factor 107 0.95 153 0.94 215 0.82 341 0.97 345 0.99 457 1.03 471 0.96 475 0.99 538 1.01 556 1.05
Experiment Mode frequency Loss factor 109 0.98 155 0.96 216 0.84 341 1.05 348 1.01 456 1.05 473 0.99 476 1.15 540 1.05 559 1.07
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Table 11. Comparison of the modal loss factors deduced from modelling and the ones obtained by experimental investigation, in the case of the structure constituted of the serge laminate with interleaved viscoelastic layer
mode 1 mode 2 mode 3 mode 4 mode 5 mode 6 mode 7 mode 8 mode 9 mode 10
Modelling Mode frequency Loss factor 108 5.74 153 2.80 216 9.74 343 14.3 348 11.3 456 12.1 471 7.39 477 9.75 540 9.41 562 10.4
Experiment Mode frequency Loss factor 110 5.80 155 2.88 215 9.65 345 14.4 351 11.1 455 12.4 474 7.52 480 9.65 543 9.45 565 10.7
The loss factors of the modes were evaluated by applying the modelling developed in Section 7 to the structure considered. The results obtained for the damping are reported in Tables 6 to 8, for the three different materials. The modal loss factors were also deduced from experimental investigation where the responses of the structure were identified in the frequency domain using MATLAB Toolbox. The results are compared in Tables 10 to 12 for the first ten modes. Also, tables report the frequencies of the free natural modes of the structure deduced from experiment and finite element analysis. A good agreement is observed between the results derived from modelling and the experimental results. Interleaving viscoelastic layer does not change significantly the frequency of the modes. Compared to the damping of the structure constituted of serge laminate, the damping of the first two modes is increased by a factor of about 5 when the structure is constituted of the sandwich material. For the other modes, the damping is increased by a factor of 1.5 to 2. In the case of the structure constituted of the serge laminate with interleaved viscoelastic layer, the damping of mode 2 (a twisting mode) is lower than the structure with sandwich material. The damping of
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
131
the other modes is greatly increased, by a factor 6 to 12 with respect to the structure constituted of the serge laminate. 40
Amplitude ( dB )
30 20 10 0 Finite element analysis Experimental results
-10 -20 0
50
100
150
200 250 300 Frequency ( Hz )
350
400
450
500
40
(a)
Amplitude ( dB )
30 20 10 0
Finite element analysis
-10
Experimental results 0
40
50
100
150
200 250 300 Frequency ( Hz )
350
400
450
500
(b)
30 Amplitude ( dB )
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
-20
20 10 0 Finite element analysis Experimental results
-10 -20 0
100
200
300
400 500 600 700 Frequency ( Hz )
800
900 1000
(c)
Figure 30. Frequency responses of the structure constituted of three different materials: (a) glass serge laminate, (b) glass serge laminate with interleaved viscoelastic layer and (c) sandwich material.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
132
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al. Table 12. Comparison of the modal loss factors deduced from modelling and the ones obtained by experimental investigation, in the case of the structure constituted of the sandwich material
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
mode 1 mode 2 mode 3 mode 4 mode 5 mode 6 mode 7 mode 8 mode 9 mode 10
Modelling Mode frequency Loss factor 109 4.25 148 4.99 423 1.89 545 1.69 588 1.64 630 1.62 767 1.56 830 1.54 853 1.53 955 1.50
Experiment Mode frequency Loss factor 110 4.30 150 4.80 425 1.95 543 1.82 590 1.70 633 1.68 769 1.70 834 1.60 855 1.63 958 1.58
Next, the modal responses of the structure were derived by finite element analysis using a mode superposition method [36]. This analysis considers the modal loss factors obtained previously and the analysis was nonlinear so as to take into account the variation of the moduli of the materials with the frequency. Figure 30 compares the frequency responses of the structure constituted of the different materials derived from the finite element analysis with the frequency responses obtained by the experimental investigation. For these responses the impact point and the measuring point considered are reported in Figure 28. The modal responses derived from finite element analysis were adjusted so as to have the amplitude response equal to zero for the frequency equal to zero. Next the experimental responses were fitted so as to have the same amplitude of the finite element analysis response and the experimental response for the first peak. The responses are reported with the same scale for the response amplitude. Due to the mode shapes and the positions of the impact and measuring points (Figure 28), the vibration modes 1, 5 and 6 are not detected in the case where the structure is constituted of serge laminate or serge laminate with interleaved viscoelastic layer. Modes 2 and 3 combine to yield two resonance peaks at frequencies of 153 Hz and 215 Hz, and an antiresonance peak at 180 Hz. Mode 4 leads to a resonance peak at 341 Hz, and modes 7 and 8 to a resonance peak about 475 Hz. In the case of the structure constituted of the sandwich material, the vibration modes 2, 3 and 5 are not detected. Modes 1, 4, 6 and 7 yield resonance peaks of 109, 545, 588 and 767 Hz, respectively. The amplitudes of the peaks are slightly decreased in the case of the structure constituted of the sandwich material. However, the higher damping is obtained in the case of the structure constituted of the serge laminate with interleaved viscoelastic layer. A significantly higher damping could be obtained using a thicker viscoelastic layer. In fact, the purpose of this section was to show that the modelling considered, associated to the experimental characterisation of the dynamic properties of the constituents, was well suited to the analysis of the damped response of a structure constituted of different composite
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Damping in Composite Materials and Structures
133
materials. The agreement between the experimental dynamic responses and the responses deduced from the modelling corroborates this ability.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Conclusions Damping properties were analysed in the case of unidirectional composites, orthotropic composites, laminates, as well as in the case of interleaved laminate materials and sandwich materials. Modelling of the damping properties of composite materials was developed considering the first-order laminate theory including the effects of the transverse shear and using the concept of the absorption of the energy induced by damping. In the case of simple structures (beams and plates), modelling has been implemented introducing the analysis of the vibrations by the Ritz method. In the case of structures of complex shape, the damping evaluation has been implemented using finite element analysis. The analysis allows us to derive the different strain energies stored in the material directions of the constituents of composite materials, and next, the energy dissipated by damping in the materials and the composite structure can be obtained as a function of the strain energies and the damping coefficients associated to the different energies stored in the material directions. Modelling so considered can be applied to any structures made of laminates, laminates with interleaved viscoelastic layers, as well as sandwich materials. Damping characteristics of laminates were evaluated experimentally using beam specimens subjected to an impulse input. Loss factors were then derived by fitting the experimental Fourier responses with the analytical motion responses expressed in modal coordinates. The damping characteristics of the composite materials and of the constituents can be deduced by applying modelling to the flexural vibrations of free-clamped beams. So it can be obtained: the loss factors in the material directions of the different layers of laminated materials, the damping characteristics of the viscoelastic layers, as well as the ones of the foam cores. The analysis has to be implemented as a function of the frequency because of the variations with the frequency of the moduli and of the damping properties of the constituents. Next, modelling associated with the damping properties obtained by the previous experimental procedure can be applied to evaluate the damping properties of any structure constituted of laminates, laminates with interleaved viscoelastic layers or sandwich materials. Then, the dynamic responses of structures can be derived by using a nonlinear mode superposition method. This procedure was applied to a simple shape structure, and the comparison between the experimental results and the results derived from modelling showed that the procedure developed is well suited to the description of the experimental results obtained.
References [1] Gibson, R. F. & Plunkett, R. A. (1977). Dynamic stiffness and damping of fiberreinforced composite materials. The Shock and Vibration Digest, 9(2), 9-17. [2] Gibson, R. F. & Wilson, D. G. (1979). Dynamic mechanical properties of fiberreinforced composite materials. The Shock and Vibration Digest, 11(10), 3-11.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
134
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
[3] Gross, B. (1953). Mathematical Structure of the Theories of Viscoelasticity. Hermann, Paris. [4] Hashin, Z. (1970). Complex moduli of viscoelastic composites- I. General theory and application to particulate composites. International Journal of Solids Structures, 6, 539-552. [5] Hashin, Z. (1970). Complex moduli of viscoelastic composites- II. Fiber reinforced materials. International Journal of Solids Structures, 6, 797-807. [6] Sun, C. T., Wu, J. K. & Gibson, R. F. (1987). Prediction of material damping of laminated polymer matrix composites. Journal of Materials Science, 22, 1006-1012. [7] Crane, R. M. & Gillespie, J. W. (1992). Analytical model for prediction of the damping loss factor of composite materials. Polymer Composites, 13(3), 448-453. [8] Adams, R. D. & Bacon, D. G. C. (1973). Effect of fiber orientation and laminate geometry on the dynamic properties of CFRP. Journal of Composite Materials, 7, 402-408. [9] Yim, J. H. (1999). A damping analysis of composite laminates using the closed form expression for the basic damping of Poisson's ratio. Composite Structures, 46, 405-411. [10] Berthelot, J. M. & Sefrani, Y. (2004). Damping analysis of unidirectional glass and Kevlar fibre composites. Composites Science and Technology, 64, 1261-1278. [11] Ni, R. G. & Adams, R. D. (1984). The damping and dynamic moduli of symmetric laminated composite beams. Theoretical and experimental results. Composites Science and Technology, 18, 104-121. [12] Adams, R. D. & Maheri, M. R. (1994). Dynamic flexural properties of anisotropic fibrous composite beams. Composites Science and Technology, 50, 497-514. [13] Lin, D. X., Ni, R. & Adams, R. D. (1984). Prediction and measurement of the vibrational parameters of carbon and glass-fibre reinforced plastic plates. Journal of Composite Materials, 18, 132-152. [14] Maheri, M. R. & Adams, R. D. (1995). Finite element prediction of modal response of damped layered composite panels. Composites Science and Technology, 55, 13-23. [15] Yim, J. H. & Jang, B. Z. (1999). An analytical method for prediction of the damping in symmetric balanced laminates composites. Polymer Composites, 20(2), 192-199. [16] Yim, J. H. & Gillespie, J. r J. W. (2000). Damping characteristics of 0° and 90° AS4/3501-6 unidirectional laminates including the transverse shear effect. Composites Structures, 50, 217-225. [17] Young, D. (1950). Vibration of rectangular plates by the Ritz method. Journal of Applied Mechanics, 17, 448-453. [18] Berthelot, J. M. (2006). Damping analysis of laminated beams and plates using the Ritz method, Composite Structures, 74(2), 186-201. [19] Ungar, E. E. & Kervin, E. M. (1962). Loss factors of viscoelastic systems in terms of energy concepts. Journal of Acoustical Society of America, 34(7), 954-957. [20] Berthelot, J. M. (1999). Composite Materials. Mechanical Behavior and Structural Analysis. Springer, New York. [21] Berthelot, J. M. (2007). Mechanical Behaviour of Composite Materials and Structures. Available on www.compomechasia.fr., Le Mans, France. [22] Hashin, Z. (1965). On elastic behaviour of fiber reinforced materials of arbitrary transverse plane geometry. Jal Mech. Phys. Solids, 13, 119. [23] Hashin, Z. (1966). Viscoelastic fiber reinforced materials., A.I.A.A. Jal, 4, 14111.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Damping in Composite Materials and Structures
135
[24] Hill, R. (1964). Theory of mechanical properties of fiber-strengthened material: I. Elastic behaviour. Jal Mech. Phys. Solids, 12, 199. [25] Christensen, R. M. (1979). Mechanics of Composite Materials, Wiley, New York. [26] Christensen, R. M. & Lo, K. H. (1979). Solutions for effective shear properties in three phase sphere and cylinder, Jal Mech. Phys. Solids, 27(4), 4. [27] Adams, R. D., Fox, M. A. O., Flood, R. J. L., Friend, R. J. & Herwitt, R. L. (1969). The dynamic properties of unidirectional carbon and glass fiber reinforced plastics in torsion and flexure. Journal of Composite Materials, 3, 594-603. [28] Adams, R. D. (1987). Damping properties analysis of composites, in Enginee-ring Handbook, Composites ASM, 1, 206-217. [29] Hwang, S. J. & Gibson, R. F. (1987). Micromechanical modelling of damping in discontinuous fiber composites using a strain energy/finite element approach. Journal of Engineering Materials and Technology, 109, 47-52. [30] Suarez, S. A., Gibson, R. F., Sun, C. T. & Chaturvedi, S. K. (1986). The influence of fiber length and fiber orientation on damping and stiffness of fiber reinforced polymer composites. Experiment Mechanics, 26(2), 175-184. [31] Hwang, S. J. & Gibson, R. F. (1992). The use of strain energy-based finite element techniques in the analysis of various aspects of damping of composite materials and structures. Journal of Composite Materials, 26(17), 2585-2605. [32] Yim, J. H. (1999). A damping analysis of composite materials using the closed form expression for the basic damping of Poisson’s ratio. Composite Structures, 46, 405-411. [33] Berthelot, J. M. & Sefrani, Y. (2007). Longitudinal and transverse damping of unidirectional fibre composites. Composites and Structures, 79(3), 423-431. [34] Young, D. (1950). Vibration of rectangular plates by the Ritz method. Journal of Applied Mechanics, 17, 448-453. [35] Timoshenko, S., Young, D. H. & Weaver, W. (1974). Vibration Problems in Engineering. John Wiley & sons, New York. [36] Berthelot, J. M. (2007). Dynamics of Composite Material and Structures. Available on www.compomechasia.fr, Le Mans, France. [37] Saravanos, D. A. & Pereira, J. M. (1992). Effects of interply damping layers on the dynamic characteristics of composite plates. AIAA Journal, 30(12), 2906- 2913. [38] Liao, F. S., Su, A. C. & Hsu, T. C. J. (1994). Vibration damping of interleaved carbon fiber-epoxy composite beams. Journal of Composite Materials, 28(8), 1840-1854. [39] Liao, F. S. & Hsu, T. C. J. (1992). Prediction of vibration damping properties of polymer-laminated steel sheet using time-temperature superposition principle. Journal of Applied Polymer Science, 45, 893-900. [40] Shen, I. Y. (1994). Hybrid damping through intelligent constrained layer treat-ments. Journal of Vibration and Acoustics, 116, 341-349. [41] Cupial, P. & Niziol, J. (1995). Vibration and damping analysis of a three-layered composite plate with a viscoelastic mid-layer. Journal of Sound and Vibration, 183(1), 99-114. [42] Yim, J. H., Cho, S. Y., Seo, Y. J. & Jang, B. Z. (2003). A study on material damping of 0° laminated composite sandwich cantilever beams with a viscoelastic layer. Composite Structures, 60, 367-374.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
136
Jean-Marie Berthelot, Mustapha Assarar, Youssef Sefrani et al.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
[43] Plagianakos, T. S. & Saravanos, D. A. (2004). High-order layerwise mechanics and finite element for the damped dynamic characteristics of sandwich composite beams. International Journal of Solids and Structures, 41, 6853-6871. [44] Saravanos, D. A. (1993). Analysis of passive damping in thick composite struc- tures. AIAA Journal, 31(8), 1503-1510. [45] Saravanos, D. A. (1994). Integrated damping mechanics for thick composite laminates. Journal of Applied Mechanics, 61(2), 375-383. [46] Berthelot, J. M. (2006). Damping analysis of orthotropic composites with interleaved viscoelastic layers: Modeling. Journal of Composite Materials, 40(21), 1889-1909. [47] Berthelot, J. M. & Sefrani, Y. (2006). Damping analysis of orthotropic composites with interleaved viscoelastic layers: Experimental investigation and discus-sion. Journal of Composite Materials, 40(21), 1911-1932. [48] Standard test method for measuring vibration damping properties of materials. 2004. ASTM E 756-04e1. Book of Standards volume 04.06. [49] Berthelot, J. M., Assarar, M., Sefrani, Y. & El Mahi, A. (2009). Damping analysis of composite materials and structures. Composite Structures, 85(3), 189-204.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
In: Composite Materials in Engineering Structures Editor: Jennifer M. Davis, pp. 137-190
ISBN: 978-1-61728-857-9 © 2011 Nova Science Publishers, Inc.
Chapter 3
MECHANICAL STATES INDUCED BY MOISTURE DIFFUSION IN ORGANIC MATRIX COMPOSITES: COUPLED SCALE TRANSITION MODELS F. Jacquemin* and S. Fréour Institut de Recherche en Génie-Civil et Mécanique (UMR-CNRS 6183), Université de Nantes – Centrale Nantes, 37 Bd de l’Université, BP406, 44602 Saint-Nazaire Cedex, France
Abstract
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Composite structures are often submitted to hygroscopic loads during their service life. Moisture uptake generates multi-scale internal stresses, the knowledge of which, granted by dedicated scale transition approaches, is precious for sizing mechanical part or predicting their durability. Experiments report that the diffusion properties of penetrant-organic matrix composite systems may continuously change during the diffusion process, due to the evolution of the internal strains experienced by the polymer matrix. On the one hand, both the diffusion coefficient and the maximum moisture absorption capacity, i.e. the main penetrant transport factors, are affected by the distribution of the local strains within the composite structure. On the other hand, accounting for strain dependent diffusion parameters change the moisture content profiles, which affect the mechanical states distribution itself. Consequently, a strong two-ways hygro-mechanical coupling occurs in organic matrix composites. The literature also reports that the effective stiffness tensor of composite plies is directly linked to their moisture content. Actually, the main parameters controlling the diffusion process remain unchanged. Thus, only the time- and depth- dependent mechanical states are affected. Consequently, this effect, independently handled, constitutes a single-way hygromechanical coupling by comparison with the above described phenomenon. This work investigates the consequences of accounting for such coupling in the modelling of the hygromechanical behaviour of composites structures through scale transitions approaches. The first part of this paper deals with the effects related to the moisture content dependent evolution of the hygro-elastic properties of composite plies on the in-depth stress states predicted during the transient stage of the diffusion process. The numerical simulations show that accounting (or not) for the softening of the materials properties occurring in practice, *
E-mail address: [email protected], [email protected]. (Corresponding author)
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
138
F. Jacquemin and S. Fréour yields significant discrepancies of the predicted multi-scale stress states. In a second part, the free-volume theory is introduced in the multi-scale hygro-mechanical model in order to achieve the coupling between the mechanical states experienced by the organic matrix and its diffusion controlling parameters. Various numerical practical cases are considered: the effect of the internal swelling strains on the time- and depth-dependent diffusion coefficient, maximum moisture absorption capacity, moisture content and internal stresses states are studied and discussed. Homogenization relations are required for estimating macroscopic diffusion coefficients from those of the plies constituents. In the third part, effective diffusivities of composite plies are estimated from the solving of unit cell problems over representative volume elements submitted to macroscopic moisture gradients when accounting for the resulting mechanical states profiles.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
1. Introduction High performance composites are being increasingly used in aerospace and marine structural applications. Organic matrix composites are often submitted to moisture and temperature environments. These environmental effects can lead to composite degradation and consequent loss of mechanical properties (Abou Msallem et al., 2008; Davies et al., 2001; Jedidi et al., 2005; Jedidi et al., 2006). An essential feature of almost all combinations of weathering conditions is humidity, hence, topics pertinent to the question of performance in the presence of moisture are of prime importance. These topics are commonly divided into two subjects. The first relates to factors which drive moisture into the composites, namely, the penetration mechanisms. The second deals with the effects of the presence of water on the performance and durability of the composites. Actually, carbon/epoxy composites can absorb significant amount of water and exhibit heterogeneous Coefficients of Moisture Expansion (CME). The CME of the epoxy matrix are effectively strongly different from those of the carbon fibers, as shown in: (Tsai, 1987; Agbossou and Pastor, 1997, Soden et al., 1998). Moreover, the diffusion of moisture in such materials is a rather slow process, resulting in the occurrence of moisture concentration gradients within their depth, during at least the transient stage (Crank, 1975). As a consequence, local stresses take place from hygroscopic loading of composite structures which closely depend on the experienced environmental conditions, on the local intrinsic properties of the constituents and on their microstructure (the morphology of the constituents, the lay-up configuration, ..., fall in this last category of factors). Now, the knowledge of internal stresses is necessary to predict a possible damage occurrence in the material during its manufacturing process or service life. Thus, the study of the development of internal stresses due to hygro-elastic loads in composites is very important in regard to any engineering application. Several studies showed the important effects of humidity on the mechanical properties of composite materials and on their long-term behaviour (Shen and Springer, 1977). In particular, problems of chemical and physical aging and of dimensional stability (swelling) caused by internal stresses may occur. Considerable efforts have been made by researchers to study the effects of moisture and temperature, and to develop analytical models for predicting the multi-scale mechanical states occurring during both the transient stage and the permanent regime of the moisture diffusion process of fiber-reinforced laminates submitted to hygro-mechanical loads (Gopalan et al., 1985; Jacquemin and Vautrin, 2002; Fréour et al., 2005-a; Jacquemin et al., 2006). In the previously cited references, the moisture diffusion process was assumed, to follow the linear, classical, and established for a
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Mechanical States Induced by Moisture Diffusion …
139
long time, Fickian law. Moreover, the materials properties involved in the hygro-mechanical constitutive behaviour required for predicting the internal stresses states were assumed independent from the moisture content distribution in these articles, whereas it is reported in the literature that moisture diffusion in composite structures entails a significant softening of the elastic properties of the composite plies (Patel et al., 2002). This phenomenon will be referred, in the following of the present work, to a weak hygro-mechanical coupling. Besides, some valuable experimental results, already reported in (Gillat and Broutman, 1978), have shown that certain anomalies in the moisture sorption process, (i.e. discrepancies from the expected Fickian behaviour) could be explained from basic principles of irreversible thermodynamics, by a strong coupling between the moisture transport in polymers and the local stress state (Weitsman, 1990-a; Weitsman, 1990-b). The present work is a synthesis of a research project dedicated to the determination of the multiscale internal stresses in the constituents of carbon-fiber reinforced epoxy composites structures submitted to hygroscopic loading, during the transient part of the moisture diffusion process, while accounting for multiple features of hygro-mechanical coupling expected to occur in practice, according to the literature. The second section of this communication is especially focused on investigating the effects related to the evolution of the elastic stiffness and coefficient of moisture expansion of the epoxy, as a function of its moisture content. The constituents properties dependence on the moisture content is determined from the evolution of the corresponding macroscopic properties, experienced in practice by the composite ply, during the transient part of the diffusion process. The required identification procedure involves an inverse self-consistent hygro-elastic scale transition model, which is described in the first paragraph of section 2. In the second paragraph of the same section, a multi-scale analysis of the transient hygro-mechanical behaviour of various composite structures submitted to hygroscopic loads is achieved. The approach entails using continuum mechanics formalism in order to perform the determination of the macroscopic mechanical states as a function of time and space, during the transient phase of the moisture diffusion process. The mechanical states of stresses and strains experienced by the constituents of each ply of the structure are determined as a function of space and time, through analytical scale transition relations. In the third paragraph of section 2, the mechanical states, predicted by the model accounting of moisture dependent properties are compared to the reference values obtained assuming the materials properties independent from the moisture content. The effect of external loading on moisture penetration into a composite material is actually a markedly relevant issue since it is difficult to imagine any application of composite materials (even non-structural) which does not result in subjecting them to some form of static or dynamic loading. It has been claimed that the general effect of such loading is to enhance the moisture-penetration mechanisms producing higher rates and maximum levels of moisture penetration. As a result, aging mechanisms taking place under the effect of moisture are also enhanced, thus decreasing the durability of the material. In the third section of the present study, a scale transition analysis, based on the free volume theory, is achieved in order to investigate the couple effect of both the external and the internal mechanical states (stresses and strains) experienced by epoxy resin composite structures, on the moisture penetration process. This work is focused on the main penetration mechanism known in composite materials, namely, the diffusion into the bulk resin matrix. The model disregards the effect of external loading on damage dependent
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
140
F. Jacquemin and S. Fréour
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
mechanisms of capillary flow along to fiber-matrix interface and flow of moisture in microflaws occurring in the matrix. The hygro-mechanical model enables to account for the proper evolution of the water diffusion behaviour parameters (rate of moisture absorption, diffusion coefficients and maximum moisture content) of the epoxy resin constituting each ply of the composite structure, during the transient part of the diffusion process. The scale transition approach provides relations linking the plies stresses (strains) to those of the constituents (epoxy and reinforcements) and homogenization procedures enabling to estimate the evolution of the plies diffusion behaviour law, from that of the epoxy, at any step of the moisture diffusion. The present work underlines the effects induced by typical laminates lay-up configurations, whose diffusion behaviour under stresses is compared to that of unidirectional composites. The numerical results, obtained according to the hygromechanical coupled model are also compared to those provided by the traditional uncoupled model. In order to predict the long-term durability of polymer matrix composite materials submitted to humid environments, the moisture diffusion behaviour has to be investigated. The knowledge of the effective diffusivity is actually required, for estimating the moisture content of polymer based fibre reinforced materials, even when a basic behaviour such as Fick’s law is assumed to occur. The scale transition relations, available in the literature, enabling to deduce the macroscopic coefficient of diffusion of a composite ply from the knowledge of its microstructure and its constituents properties, were established assuming a stress/strain free state within the ply and its constituents (Hashin, 1972). The purpose of the fourth section of this paper is to propose original analytical solutions for the effective moisture diffusivity of organic matrix composite materials, through appropriate hygromechanical scale-transition models. The effective diffusivity is deduced from solving a unit cell problem on a Representative Volume Element (RVE) on which is imposed an average macroscopic strain or an average macroscopic stress.
2. Effects of Moisture Dependent Constituents Properties on the Hygroscopic Stresses Experienced by Composite Structures 2.1. Inverse Scale Transition Modelling for the Identification of the Hygro-Elastic Properties of One Constituent of a Composite Ply 2.1.1. Introduction The precise knowledge of the local properties of each constituent of a composite structure is required in order to achieve the prediction of its behavior (and especially its mechanical states) through scale transition models. Nevertheless, the stiffness and coefficients of moisture expansion of the matrix and reinforcements are not always fully available in the already published literature. The practical determination of the hygro-mechanical properties of composite materials are most of the time achieved on uni-directionnaly reinforced composites whereas the properties of the unreinforced matrices are easily accessible though measurements, too (Bowles et al., 1981; Dyer et al., 1992; Ferreira et al., 2006-a; Ferreira et
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
141
al., 2006-b; Herakovich, 1998; Sims et al., 1997). In spite of the existence of several articles dedicated to the characterisation of the properties of the isolated reinforcements (DiCarlo, 1986; Tsai and Daniel, 1993; Tsai and Chiang, 2000), the practical achievement of this task remains difficult to handle, and the available published data for typical reinforcing particulates employed in composite design are still very limited. As a consequence, the properties of the single reinforcements exhibiting extreme morphologies (such as fibers), are not often known from direct experiment, but more usually they are deduced from the knowledge of the properties of the pure matrices and those of the composite ply (which both are easier to determine), through appropriate calculation procedures. In the present case, the literature provides the moisture dependent evolution of uni-directional fiber-reinforced plies elastic moduli (Patel et al., 2002), but not the corresponding properties for the constituents. Thus, a dedicated identification method is necessary before proceeding further. The question of determining the properties of some constituents of heterogeneous materials has been extensively addressed in the field of materials science, especially for studying complex polycrystalline metallic alloys, like titanium alloys, (Fréour et al., 2002; Fréour et al., 2005-b; Fréour et al., 2006) or metal matrix composites (typically Aluminum-Silicon Carbide composites (Fréour et al., 2003-a; Fréour et al., 2003-b) or iron oxides from the inner core of the Earth (Matthies et al., 2001, for instance). The required calculation methods involved in order to achieve such a goal are either based on Finite Element Analysis (Han et al., 1995) or on the inversion of scale transition homogenization procedures (Fréour et al., 2002; Fréour et al., 2003-a; Fréour et al., 2003-b; Fréour et al., 2005-b; Fréour et al., 2006): this solution will be extensively used in the following of the present work. Numerical inversion of EshelbyKröner hygro-elastic self-consistent model will be summarized and discussed in the following of this very section.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
2.1.2. Estimating Constituents Properties from Eshelby-Kröner Self-consistent Inverse Scale Transition Model 2.1.2.1. Introduction Scale transition models are based on a multi-scale representation of materials. In the case of composite materials, for instance, a two-scale model is sufficient: -
-
The properties and mechanical states of either the resin or the reinforcements are respectively indicated by the superscripts m and r. These constituents define the so-called “pseudo-macroscopic” scale of the material. Homogenisation operations performed over its aforementioned constituents are assumed to provide the effective behaviour of the composite ply, which defines the macroscopic scale of the model. It is denoted by the superscript I.
2.1.2.2. Estimating the Effective Properties of a Composite Ply through Eshelby-Kröner Self-consistent Model Within scale transition modelling, the local properties of the i−superscripted constituents are usually considered to be known (i.e. the pseudo-macroscopic stiffnesses, Li and coefficients of moisture expansion βi), whereas the corresponding effective macroscopic
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
142
F. Jacquemin and S. Fréour
properties of the composite structure (respectively, LI and βI) are a priori unknown and results from (often numerical) computations. Among the numerous, available in the literature scale transition models, able to handle such a problem, most involve rough-and-ready theoretical frameworks: Voigt (1928), Reuss (1929), Neerfeld-Hill (Neerfeld, 1942; Hill, 1952), Tsai-Hahn (Tsai and Hahn, 1980), and Mori-Tanaka (Tanaka and Mori, 1970; Mori and Tanaka, 1973) approximates fall in this category. This is not satisfying, since such a model does not properly depict the real physical conditions experienced in practice by the material. In the field of scale transition modelling, the best candidate remains Kröner-Eshelby self-consistent model (Eshelby, 1957; Kröner, 1958), because only this model takes into account a rigorous treatment of the thermo-hygroelastic interactions between the homogeneous macroscopic medium and its heterogeneous constituents, as well as this model enables handling the microstructure (i.e. the particular morphology of the constituents, especially that of the reinforcements). The method was initially introduced to treat the case of polycrystalline materials in pure elasticity. The model was thereafter extended to thermo-elastic loads and gave satisfactory results on either singlephase or two-phases materials (Fréour et al., 2003-a; Fréour et al., 2003-b). More recently, this classical model was improved in order to treat hygroscopic load related questions. Therefore, the formalism was extent so that homogenisation relations were established for estimating the macroscopic coefficients of moisture expansion (Jacquemin et al., 2005). The main equations involved in the determination of the effective hygro-elastic properties of heterogeneous materials through Kröner-Eshelby self-consistent approach reads:
[
(
])
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
−1 LI = Li : I + E I : Li − LI
I
β =
1 ΔC
(L + L : R ) i
I
I
I −1
I
−1
(
i
i = r, m
I
: L + L :R
:L
)
I −1
i =r,m
(1) i
i
: L : β ΔC
i i =r,m
(2)
Where ΔCi is the moisture content of the studied i element of the composite structure. The superscripts r and m are considered as replacement rules for the general superscript i, in the cases that the properties of the reinforcements or those of the matrix have to be considered, respectively. Actually, the pseudo-macroscopic moisture contents ΔCr and ΔCm can be expressed as a function of the macroscopic hygroscopic load ΔCI (Loos and Springer, 1981). In relations (1-2), the brackets < > stand for volume weighted averages. Hill (1952) suggested arithmetic or geometric averages for achieving these operations. Both have been extensively used in the field of materials science. The interested reader can refer to (Morawiec, 1989; Matthies and Humbert, 1993; Matthies et al., 1994) that take advantage of the geometric average for estimating the properties and mechanical states of polycrystals, whereas (Fréour et al., 2003-b; Jacquemin et al., 2005; Kocks et al., 1998) show applications of arithmetic average. In a recent work, the geometric average was tested for estimating the effective properties of carbon-epoxy composites (Fréour et al., 2007). Nevertheless, the obtained results were not found as satisfactory as in the previously
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
143
studied cases of metallic polycrystals or metal ceramic assemblies. Consequently, arithmetic average only will be used in the following of this manuscript. In the present case, where the macroscopic behaviour is described by two, separate, heterogeneous inclusions only (i.e. one for the matrix and one for the reinforcements), introducing vr and vm as the volume fractions of the ply constituents, and taking into account the classical relation on the summation over the volume fractions (i.e. vr + vm = 1), the volume average of any tensor A writes:
Ai
i = r,m
= vr Ar + vm Am (3)
According to equations (1-2), the effective properties expressed within Eshelby-Kröner self-consistent model involve a still undefined tensor, RI. This term is the so-called “reaction tensor” (Kocks et al., 1998). It satisfies:
(
)
I I R I = I − S esh : S esh
−1
= ⎛⎜ LI ⎝
−1
− E I ⎞⎟ : E I ⎠
−1
(4)
In the very preceding equation, I stands for the fourth order identity tensor. Hill’s tensor E , also known as Morris tensor (Morris, 1970), expresses the dependence of the reaction tensor on the morphology assumed for the matrix and its reinforcements (Hill, 1965). It can I I I −1 I . It has to be be expressed as a function of Eshelby’s tensor Sesh , through E = Sesh : L underlined that both Hill’s and Eshelby’s tensor components are functions of the macroscopic stiffness LI. Some examples are given in (Kocks et al., 1998; Mura, 1982). Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
I
2.1.2.3. Inverse Eshelby-Kröner Self-consistent Elastic Model The pseudomacroscopic stiffness tensor of the reinforcements can be deduced from the inversion of the Eshelby-Kröner main homogenization form over the constituents elastic properties (1) as follows: r
L =
1 v
r
I
[ ( I
r
I
) ]
L : E : L −L +I −
v
m
v
r
m
[ ( I
m
I
) ] : [E : (L
L : E : L −L +I
−1
I
r
I
) ]
−L +I
(5)
The application of this equation implies that both the macroscopic stiffness and the pseudomacroscopic mechanical behaviour of the matrix are perfectly determined. The elastic stiffness of the matrix constituting the composite ply will be assumed to be identical to the elastic stiffness of the pure single matrix, deduced in practice from measurements performed on bulk samples made up of pure matrix. It was demonstrated in (Fréour et al., 2002) that this assumption was not leading to significant errors in the case that polycrystalline multi-phase samples were considered.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
144
F. Jacquemin and S. Fréour
An expression, analogous to above-relation (5) can be found for the elastic stiffness of the matrix, through the following replacement rules over the superscripts/subscripts: m → r, r → m . In the particular case, where impermeable reinforcements are present in the composite r
structure, ΔC = 0 . Accounting of this additional condition, the pseudo-macroscopic coefficients of moisture expansion of the matrix can be deduced from the inversion of the homogenization form (2) as follows (an extensive study of this very question was achieved in Jacquemin et al., 2005):
β
m
=
ΔC v
m
I
ΔC
m −1
m
L
(
m
I
I
) (
i
I
: L +L :R : L +L :R
)
I −1
I
:β
:L
I
i =r,m
(6)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
2.1.2.4. Application of Inverse Scale Transition Model to the Determination of the Moisture and Temperature Dependent Pseudo-macroscopic Elastic Properties of Carbon-epoxy Composites The literature provides evolutions for the elastic properties of carbon-fiber reinforced epoxies, as a function of the moisture concentration and the temperature (Patel et al., 2002; Sai Ram and Sinha, 1991). Table 1 of the present work summarizes the previously published data for an unidirectional composite designed for aeronautic applications, containing a volume fraction vr=0.60 of reinforcing fibers. These evolutions of the macroscopic mechanical properties are obviously directly related to the variation of the pseudo-macroscopic elastic properties experienced by the composite plies constituents, as a function of the environmental conditions. Nevertheless, it is usually assumed that carbonfibers do not absorb water, thus, there is no reason for expecting to link the elastic properties of the reinforcements to the moisture content. Moreover, carbon fibers are a ceramic, and ceramics usually present thermo-mechanical properties being almost independent from temperature, contrary to metals or polymers (Fréour et al., 2003a; Fréour et al., 2003b). Furthermore, according to Table 1, the macroscopic longitudinal Young I modulus Y1 is independent from the environmental conditions, in the studied ranges of temperatures (TI comprised between 300 K and 400 K), and macroscopic moisture content (CI holds within 0 to 0.75 %), the longitudinal direction being parallel to the principal axis of the fibers. Now, it is well known that the macroscopic properties of such an unidirectional composite ply are governed by the pseudo-macroscopic properties of the reinforcements in the direction parallel to the fiber axis, whereas, on the contrary, they mainly depend on the pseudo-macroscopic properties of the constitutive matrix, along directions perpendicular to the fiber axis (see, for instance, (Herakovich, 1998; Tsai and Hahn, 1980)). As a consequence, on the basis of the values presented in Table 1, it can be reasonably considered that the elastic properties of the carbon reinforcements are independent from the environmental conditions applied to the material. Thus, in first approximation, the properties of the reinforcements will be considered as fixed, and will be I I identified once and for all. However, the decreasing of Y2 (and G 12 ), observed for
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
145
increased temperatures or moisture concentrations, according to Table 1 implies a softening of the pseudo-macroscopic elastic properties of the epoxy. Thus, the elastic moduli of the matrix should be identified for each available set of macroscopic data, in order to find their susceptibility to hygro-thermal conditions. As a consequence, due to the time-span of the moisture diffusion process, each ply of the composite structure and the constitutive epoxy matrix of them are expected to present different hygro-elastic properties from those of the neighbouring plies (and their constituting matrix), during the transient part of an hygroscopic load. The consequence of this physical phenomenon on the multi-scale stress distribution in composite structures will be studied in the next section.
2.2. Multi-Scale Stresses Estimations in Composite Structures Accounting of Hygro-Mechanical Coupling for the Elastic Stiffness: T300/5208 Composite Pipe Submitted to Environmental Conditions Thin laminated composite pipes, with 4 mm thickness, initially dry then exposed to an ambient fluid, made up of T300/5208 carbon-epoxy plies, with a fiber volume fraction vr=0.6, were considered for the determination of both macroscopic stresses and moisture contents as a function of time and space. Table 1. Experimental macroscopic elastic moduli dependent on moisture contents and temperatures, according to (Sai Ram and Sinha, 1991)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
macroscopic hygro-thermal load moisture content ΔCI [%] 0 0.25 0.75 0 0
Temperature TI [K] 300 300 300 325 400
Macroscopic elastic moduli
Y1I [GPa] 130 130 130 130 130
Y2I [GPa]
I ν12 [1]
I G12 [GPa]
G I23 [GPa]
9.5 9.25 8.75 8.5 7.0
0.3 0.3 0.3 0.3 0.3
6.0 6.0 6.0 6.0 4.75
3.0 3.0 3.0 3.0 2.39
Table 2. Pseudo-macroscopic elastic moduli and stiffness tensor components assumed for the epoxy matrix of the composite plies at ΔC I = 0 % and TI = 300 K, according to (Herakovitch, 1998)
Elastic moduli
Stiffness tensor components
Y1m [GPa]
Y2m [GPa]
m ν12 [1]
m G12 [GPa]
Gm 23 [GPa]
5.35
5.35
0.350
1.98
1.98
Lm 11 [GPa]
Lm 22 [GPa]
Lm 12 [GPa]
Lm 44 [GPa]
Lm 55 [GPa]
8.62
8.62
4.66
1.98
1.98
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
146
F. Jacquemin and S. Fréour
Figure 1. Time and space dependent moisture content profiles in the composite structure.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 2. Time-dependent profile of the macroscopic transverse coefficient of moisture expansion.
The following hygroscopic external conditions were considered: a symmetric loading corresponding to a relative humidity of 100 % on each boundary of the structure (so that the moisture content is equal to 1.5 %). The corresponding time-dependent moisture content profiles, obtained assuming that the moisture diffusion process follows Fick’s law, are depicted on Figure 1. The time-dependant evolution of the moisture content in each ply of the structure is associated to an evolution of the macroscopic and local hygro-elastic properties, according to the method proposed in section 2 of the present work. An example is given on Figure 2, for the macroscopic transverse coefficient of moisture expansion of the composite plies constituting the cylinder. The closed-form formalism used in order to determine the mechanical stresses and strains in each ply of the structure, induced by the distribution of moisture content, is described in (Jacquemin and Vautrin, 2002). The pseudo-macroscopic states of stresses and strains, experienced by the constituents of a given ply, are determined from their macroscopic counterparts (included the moisture content in the considered ply), trough the analytical scale transition relations established in (Fréour et al., 2005a) on the basis of the fundamental analytical achievements previously published in (Welzel et al., 2005). Figures 3 and 4 show the numerical results obtained for the time-dependent multi-scale distribution of transverse and shear stresses for ± 55° laminates and uni-directionnal composites, respectively (obviously for the uni-directionnaly reinforced structure, no shear stresses do
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
147
occur, so that the corresponding pictures have not been provided). Figures 3 and 4 report the results obtained for two specific plies only: the external and the central plies of the hollow cylinder.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
2.3. Discussion about the Results (i) The results of figure 1 are typical of previously published works (Jacquemin et al., 2005): the transient part of the (slow) moisture diffusion process in composite materials induces strong moisture content gradients within the depth of the structure. The strongest gradients occur at the beginning of the diffusion process and weaken as the moisture content increases in the bulk of the structure: along the time, the saturation ensures that each ply of the structure experiences the same moisture content. (ii) Figure 2 provides original additional interesting results: The numerical simulations show that strong gradients occur for the macroscopic transverse coefficient of moisture expansion, especially at the vicinity of both the external and internal plies of the studied hollow cylinder, during the transient part of the moisture diffusion process. At the contrary, the hygro-mechanical properties at any scale reach an uniform value when the permanent state is attained. Nevertheless, strong discrepancies between the macroscopic / local properties do still remain even at the saturation of the diffusion process, depending on the choice of the hypothesis concerning the dependence of the properties on the moisture content. (iii) According to figures 3 and 4, the fact of considering (or not) an evolution of the hygro-elastic properties of both the composite plies and its constitutive matrix do strongly affects the transverse stresses levels and their distributions in the plies and the constituents of them. According to Figure 3, in the laminate, only the macroscopic stresses and those of the epoxy matrix do significantly vary as a function of the hypothesis made on the materials properties: accounting of a softening of both the transverse Young’s modulus of the matrix and the ply during the moisture diffusion obviously weakens the amount of transverse stresses induced by the hygroscopic load at macroscopic scale and at pseudo-macroscopic scale in the epoxy. The predicted stress states experienced by the plies and its constitutive matrix can be reduced by up to 30% in the case that the realistic evolution of the materials properties are taken into account, by comparison with the results of the simulation performed without taking into account of that additional physical phenomenon. Figure 4 reports the classical results expected in the case that a uni-directional composite is submitted to a transient hygroscopic load: the macroscopic stresses raise at the beginning of the moisture diffusion, but decrease thereafter, so that the plies are not anymore submitted to any stress when the permanent state is reached. However, the absolute value of the corresponding pseudo-macroscopic transverse stresses increase almost continuously during the moisture diffusion process, so that the strongest stress level occur when the saturation state is attained. It should be underlined that in this specific case, the pseudomacroscopic transverse stresses calculated for the carbon-fiber vary significantly
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
148
F. Jacquemin and S. Fréour
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
depending on the choice of the hypothesis concerning the dependence of the properties on the moisture content. (iv) Macroscopic and local shear stresses are negative for the external ply, whereas they are positive for the central ply of the considered structure. According to figure 3, accounting of an evolution of the materials properties as a function of the moisture content experienced by the composite ply do have an effect on the concentration of the shear stresses within the reinforcements, contrary to the case previously studied of the transverse stresses. From the three calculated shear stresses (i.e., those of the ply, the matrix, or the reinforcements), the hypothesis of a possible evolution of the materials properties with the moisture content has its strongest effect on the reinforcements shears stress states, in spite of the fact that the carbon fibers properties are actually constant during the moisture diffusion process. The weaker local shear stresses experienced by the carbon fibers come from the localization of a weaker macroscopic counterpart, which itself is explained by the softening of the plies hygro-elastic properties as a function of the moisture content.
Figure 3. Multi-scale stress states in (a) the external ply / (b) the central ply of a ± 55° composite during the transient part of the moisture diffusion process.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
149
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
(v) According to the comments i) to iv) listed above, accounting of an evolution of the multi-scale hygro-elastic properties of composite plies has two main consequences which can be considered as responsible for the reduced amount of estimated stresses compared to the reference values (corresponding to the estimations achieved in the case that the properties of the dry material are considered to be still valid at any time during the moisture diffusion process). Firstly, strong deviations occur between the effective properties of the humid material, and their counterparts for the dry material (see figure 2). This effect increases along the time, as the mass of water having penetrated the structure increases, and reaches its maximum when the permanent stage of the diffusion process occurs. Since the predicted stresses are obviously intimately linked to the hygro-elastic properties exhibited by the material, this effect partially explains the discrepancies displayed on figures 3 and 4, between the two sets of curves (depending on the assumption considered for defining the materials properties). Secondly, moisture contents gradients occur during the transient stage of the moisture diffusion within the structure, since it is a rather slow process (see figure 1). The distribution of the hygroscopic load within depth of the structure directly induces a distribution of the hygro-elastic properties, in the case that their dependence on the moisture content is taken into account for achieving the calculations. Heterogeneous distributions of the hygro-mechanical properties explain therefore the discrepancies occurring at the beginning of the moisture diffusion process between the internal stresses predicted depending on whether the practical evolution of the materials properties as a function of the moisture content are taken into account or not. Thus, the effects on the raising of internal stresses, related to the softening of the material induced by the diffusion of water, can be expected to occur even at the beginning of the hygroscopic loading of a composite structure.
Figure 4. Multi-scale stress states in (a) the external ply / (b) the central ply of a uni-directional composite during the transient part of the moisture diffusion process.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
150
F. Jacquemin and S. Fréour Table 3. Pseudo-macroscopic elastic moduli and stiffness tensor components identified for the carbon fiber reinforcing the composite plies at ΔC I = 0 % and TI = 300 K, according to Eshelby-Kröner inverse self-consistent model. Comparison with the corresponding properties exhibited in practice by typical high strength carbon fibers, according to (Herakovitch, 1998) Elastic moduli
Eshelby-Kröner inverse model Typical expected properties Stiffness tensor components Eshelby-Kröner inverse model Typical expected properties
G r23
r G12
[GPa]
[GPa]
0.27
4.0
12.1
15
0.279
5.0
15
Lr11 [GPa]
Lr22 [GPa]
Lr12 [GPa]
Lr44 [GPa]
Lr55 [GPa]
219.2
23.9
10.8
4.0
12.1
236.7
20.1
8.4
5.0
15
Y1r [GPa]
Y2r [GPa]
r ν12 [1]
213.2
13.3
232
In the next section, the free-volume theory is introduced in the multi-scale hygromechanical model in order to achieve the coupling between the mechanical states experienced by the organic matrix and its diffusion controlling parameters.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
3. Stress-Dependent Moisture Diffusion in Composite Materials 3.1. Accounting for a Coupling between the Mechanical States and the Moisture Diffusion in Pure Organic Matrix Polymer matrix constitutes the preferential penetration path for small molecule penetrants, such as water, diffusing through organic matrix composites, especially in the cases when their reinforcements are impermeable. Dense polymer only will be considered in the present work. Thus, according to this additional condition, the occurrence of voids or porosities will be neglected, in the following. Dense polymers have no pores, however, there exists the thermally agitated motion of chain segments providing penetrant-scale transient gaps (free volume) in the polymer matrix allowing penetrants to diffuse into the bulk of the material, from one side of the structure to the other (Chen et al., 2001). The size and shape of the thermally induced cavities, available in polymers controls the rate of gas diffusion and its permeation properties (Adamson, 1980; Wang et al., 2003). These cavities and packing irregularities actually constitute the so-called “free-volume” of the material. The free-volume concept is extensively taken into account for explaining many properties presented by polymers, such as their visco-elastic behaviour (Vaughan and McPherson, 1973). Especially, transport phenomena in polymers are generally explained by theories based on the extensive involvement of the free volume notion (Crank and Park, 1968). The free volume actually corresponds to the difference between the specific (macroscopic) volume of the polymer and the actual volume occupied by its constitutive molecules. It is obvious, that the transport
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
151
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
mobility of particles in a well-packed system depends on the degree of packing of the system, m an inverse of measure of which is the matrix free volume fraction: v f , in the case that an organic matrix composite with impermeable reinforcements, is considered. Free volume concept theoretical approaches, based on the works of Cohen and Turnbull (1959) are often applied to the study of gas permeation. For instance, diffusion coefficients of gases in polymers can be deduced from the equations established by Fujita (1991). Moreover, it has also been realised, that the heterogeneous swelling, which accompanies sorption or desorption of water, leads to the creation of a multi-scale stress pattern in a composite structure. This aspect has been the subject of numerous papers over the past few years (Davies et al., 2001; Jedidi et al., 2005; Jedidi et al., 2006; Gigliotti et al., 2007). As early as 1953, Crank had suggested that the swelling stresses in polymer membranes, through which a penetrant diffuses, affects the diffusion coefficient (Crank, 1953). The question is still addressed in the recent literature (Larobina et al., 2007). As a consequence, due to the heterogeneous moisture diffusion behaviour of the epoxy resins, on the one hand, and that of the carbon fibers, on the other hand, the hygro-mechanical coupling occurring in composite structures initiates at the scale of the constituents, and especially within the organic matrix. In order to reach the goal of the present work: the modelling of the hygro-mechanical behaviour of organic matrix composite laminates, a multiscale is, thus, obviously required. However, employing such a multi-scale approach implies as usual the knowledge of the specific behaviour of the single elementary constituents of the representative elementary volume. As a consequence, the following paragraphs will be devoted to the study of the hygro-mechanical coupling existing in the single polymer matrix. According to the literature, this coupling affects both parameters governing the Fickian diffusion law: (i) the diffusion coefficient, and (ii) the maximum moisture absorption capacity. The hygro-mechanical coupled model described in the present study does not satisfy the fundamental principles of irreversible thermodynamics, contrary to the more realistic frameworks developed, on the one hand, by Aboudi and Williams (2000) and, on the other hand, by Derrien and Gilormini (2007; 2009). Both the models proposed in these papers do present significant drawbacks, constituting the reasons why they were not considered in the present work. While it can be applied for modelling the transient behaviour of multi-phase materials such as composites, Aboudi and William’s hygro-thermo-mechanical coupled approach derives from the historical works written by Sih and his collaborators (1986) and Weitsman (1987) which do not handle any source of anisotropy. Thus, their model can hardly be employed for modelling such strongly anisotropic materials as organic matrix composites structures reinforced by carbon fibers. On the contrary, (Derrien and Gilormini, 2009) achieves an investigation of the hygro-mechanical coupled effects in transversely isotropic elastic polymer-matrix composites, but this study is focused on the steady state only, i.e. on the cases when the moisture flux is null everywhere. Since it was demonstrated in (Jacquemin et al., 2005; Fréour et al., 2005-a), that the strongest internal stress states could occur during the transient stage of the moisture diffusion process, the model proposed by Derrien and
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
152
F. Jacquemin and S. Fréour
Gilormini does not provide enough information for presenting an help for designing composite parts conceived for withstanding hygro-mechanical loads during their service life. Table 4. Moisture and temperature dependent pseudo-macroscopic elastic moduli and stiffness tensor components identified for the epoxy matrix constituting the composite plies, according to Eshelby-Kröner inverse self-consistent model macroscopic hygro-thermal load moisture Temperature content ΔCI TI [K] [%] 0 300 0.25 300 0.75 300 0 325 0 400
Stiffness tensor components
Elastic moduli
Ym
[GPa] 5.35 5.22 4.95 4.81 4.17
ν m [1] 0.35 0.33 0.28 0.25 0.27
G
m
[GPa] 1.98 1.98 1.98 1.98 1.04
Lm 11
Lm 12
Lm 44
[GPa] 8.62 7.68 6.29 5.76 5.20
[GPa] 4.67 3.75 2.41 1.91 1.92
[GPa] 1.98 1.98 1.98 1.98 1.04
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
3.1.1. Moisture Diffusion Coefficient Following the experimental observations, a theoretical approach was suggested that was based on the calculation of the free-volume change in the stressed state. This work corresponded with Fahmy and Hurt’s (1980) ideas, who calculated the free-volume change under stresses for an epoxy resin. This calculation is based on the modelling of the polymer by an assembly of thick spherical shells having the same ratio of inner to outer radii. The volume fraction of the spherical cavity (being the same for all shells) represents the free volume fraction of the organic matrix. In practice, it was later observed by Neumann and Marom, that the main mechanical state controlling the hygro-mechanical coupling was the strain instead of the stresses (Neumann and Marom, 1985; Neumann and Marom, 1987). Accordingly, the following present work will develop a model based on the strains. Assuming that the Fickian diffusion coefficient was related to the free volume by the Doolittle’s equation, the authors proposed an expression for the ratio of the diffusion coefficients in the strained and free-of-strain states (Neumann and Marom, 1985 ; Neumann and Marom, 1986):
⎛ Dm Ln⎜⎜ εm ⎝ D0
⎞ a ⎛ 1 1 ⎟= m⎜ m − m ⎟ v ⎜v ⎝ f0 v fε ⎠
⎞ ⎟ ⎟ ⎠
(7)
m Where D 0 and D εm are the Fickian moisture diffusion coefficients for the strain-free matrix and that of the strained epoxy, respectively, whereas a is an empirically deduced factor. m
m
vm denotes the volume fraction of the matrix in the composite ply, v f0 and v fε are the free volume fraction of the strain-free epoxy and that of the strained organic matrix, respectively.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
153
m
Many authors agree that the value of v f0 for a defect-free resin tends towards 2.5% (Fahmy and Hurt, 1980; Neumann and Marom, 1986). The free-volume fraction for a strained epoxy is related to its counterpart existing in the corresponding unstrained resin through: m v fε
=
m v f0
+
ΔV m V0m
(8)
m Where ΔV stands for the organic matrix volume variation induced by the strains. It V0m corresponds in practice to the trace of the strain tensor experienced by the polymer matrix
Tr ε
m
:
ΔV m Vεm − V0m = = Tr ε m m m V0 V0 m
Where Vε
m
and V0
(9)
denote the volume of the strain organic matrix and that of the
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
unstrained epoxy, respectively. Equations (7) and (8) yield the following expression for the Fickian moisture diffusion coefficients of the strained/unstrained resins:
ΔV m ⎛ Dεm ⎞ a a Tr ε m V0m = m Ln⎜⎜ m ⎟⎟ = m m m m m ⎝ D0 ⎠ v v m ⎡ v m + ΔV ⎤ v v f0 v f0 + Tr ε f0 ⎢ f0 V0m ⎥⎦ ⎣
[
] (10)
In order to estimate the diffusion coefficient of the strained polymeric matrix D εm , the knowledge of both the organic matrix strain tensor and its strain-free diffusion coefficient
D 0m is required. D 0m can be deduced from the slope of the mass uptake evolution as a function of the time power ½, when the time tends towards 0 (thus, at the beginning of the diffusion process). In the case that a N5208 epoxy is considered, the value of this coefficient, measured at room temperature is: D 0m = 13,484 × 10 −8 mm 2 s −1 . According to equation (10), amorphous polymer matrices experiencing any kind of strain state are subjected to a variation of their free-volume. Such a mechanical state, governing free-volume change, involves an evolution of the water-like, small molecule penetrants, diffusion law parameters. The moisture diffusion coefficient increases when the trace of the strain tensor experienced by the polymer is positive, and decreases when this trace is negative. This formalism is consistent with the practical observations.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
154
F. Jacquemin and S. Fréour
One observes actually in practice, that the diffusion coefficient of a pure epoxy matrix increases when it is submitted to an uniaxial tensile load, whereas this coefficient decreases when the polymer is submitted to an uniaxial compressive load (Fahmy and Hurt, 1980). The same authors state, in the same article, that the increased water uptake observed after submitting a polymer to tensile load is dependent not only on the diffusion coefficient but also on the equilibrium water concentration and that the stress state may indeed influence both of these parameters. According to (Fahmy and Hurt, 1980) the hygro-mechanical coupling has not only an influence over the evolution of the diffusion coefficient, but on the maximum moisture absorption capacity, too. This second feature will be developed extensively in the next paragraph.
3.1.2. Maximum Moisture Absorption Capacity The maximum moisture absorption capacity for an unstrained epoxy satisfies (see for instance Neumann and Marom, 1986):
M ∞m0 = v f0m ×
ρw ρm
(11)
Let us assume that the maximum moisture absorption capacity of the same organic matrix experiencing any strain state becomes:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
M ∞mε = v fmε ×
ρw ρm
(12)
Equations (11) and (12) yield:
(
)
ρw m m m m M∞ − M = v − v × ε ∞0 fε f0 ρm
(13)
Combining relations (8) and (13) leads to:
M ∞mε = M ∞m0 +
ΔV m ρ w × V0m ρ m
(14)
According to the experimental results presented, for instance, by Neumann and Marom, (Neumann et Marom, 1987), on pure epoxies, parameter a, appearing in equation (9), especially, does vary in a narrow range: 0.031 ≤ a ≤ 0.036 with an average value equal to 0.033.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
155
3.2. Composite Materials 3.2.1. Modelling the Moisture Diffusion Process Consider a laminated hollow cylinder, whose inner an outer radii are respectively a and b, composed of n plies delimitated by cylinders with radii ri and ri+1. To solve the Fick’s equation, the boundary moisture contents have to be known. The moisture content at saturation is determined from the moisture content of the matrix (14). The mixed law on the volumes reads: I
f
f
m
V = v ×V + v ×V
m
(15)
Where VI, Vf and Vm are, respectively, the volumes of composite, fibers and matrix. vf and vm are, respectively, the volume fractions of the fibers and matrix. Equation (15) can be expressed by introducing the moisture contents: I f f f m m m ρI × M∞ ε = v × ρ × M ∞ε + v × ρ × M ∞ε
(16)
Where ρI, ρf and ρm are, respectively, the densities of the composite, fibers and matrix. Since, the fibers do not absorb water, equation (16) is simplified as follows:
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
I m M∞ ε =v
ρ
m
ρI
m M∞ ε
(17)
The macroscopic moisture content (at ply scale) is a solution of the Fick’s equation with a moisture diffusion coefficient dependent on the matrix one, which is dependent on the local mechanical state. The expression of the effective moisture diffusion coefficient as a function of the moisture diffusion coefficient of the matrix is (Hashin, 1972).
D εI = D εm
1 − vf 1 + vf
(18)
The moisture diffusion coefficient of the matrix is determined through equation (9) from the local strains deduced from the localization of the macroscopic strains. The hygromechanical coupling induces for the constitutive plies different moisture diffusion coefficients and moisture contents. Thus, the moisture flux is continuous at the interply and the moisture content is discontinuous (Jacquemin et al., 2002). The Fickian problem (19-20) can be expressed as:
⎡ ∂ 2 C i 1 ∂C i ⎤ ∂C i = Di (t )⎢ + ⎥ a 1 , the diffusion coefficient increases as shown in Figure 16. On the other hand, we have:
ψε = αε
D fε D εm
(67)
Thus:
ψ ε α ε D fε D 0m = ψ 0 α 0 D f0 D εm f
(68)
f
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Assuming that D ε ≈ D 0 , expression (68) simplifies as follows:
α ε D 0m ψε = ψ0 α 0 D εm
(69)
Considering (69) and (65), we arrive at the following expression:
⎡ ΔV m ⎢ ⎢ V0m α ψ ε = ψ 0 ε exp ⎢- a α0 ⎛ ΔV m ⎢ m⎜ m v v + f0 ⎜ f0 ⎢ V0m ⎝ ⎣
⎤ ⎥ ⎥ ⎥ ⎞⎥ ⎟ ⎟⎥ ⎠⎦
(70)
Since the variation of α ε as a function of the strain is almost negligible, whatever the fiber volume fraction, the expression (70) becomes:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
178
F. Jacquemin and S. Fréour
⎡ ΔV m ⎢ ⎢ V0m ψε = exp ⎢- a ψ0 ⎛ ΔV m ⎢ m⎜ m v v + f0 ⎜ f0 ⎢ V0m ⎝ ⎣
⎤ ⎥ ⎥ 1 ⎥= ⎞ ⎥ δ Dooli ⎟ ⎟⎥ ⎠⎦
(71)
Finally, one gets: ΔV m ⎛ψ ln⎜⎜ ε ⎝ ψ0
⎞ ⎟⎟ = -a ⎠
V0m ⎛ ΔV m⎜ m v f0 v + ⎜ f0 V0m ⎝
m
On Figure 17, the ratio
⎛ Dm = − ln⎜ εm ⎜D ⎞ ⎝ 0 ⎟ ⎟ ⎠
⎞ ⎛ m ⎟ = ln⎜ D 0 ⎟ ⎜ Dm ⎠ ⎝ ε
⎞ ⎛ eff ⎟ = ln⎜ D 0 ⎟ ⎜ D eff ⎠ ⎝ ε
⎞ ⎛ ⎟ = a⎜ 1 − 1 ⎟ ⎜ vm vm f0 ⎠ ⎝ fε
⎞ ⎟ ⎟ ⎠
(72)
ψε is plotted for different stress states and reinforcements ψ0
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
volume fractions. We note that the ratio is sensitive only to the mechanical loading. This ratio is greater than one for the traction and less than one for compression. The question that arises now is to express the effective diffusivity of the RVE experiencing strains according to the permeability index ψ ε .
Figure 17. Evolution of
ψε ψ0
for different stress states.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
179
By considering, the expression (71) and (66), we arrive to the following expression:
D εeff − per D εm
1 − vf f = 1+ v
+ δ Dooli ψ ε
1 + ψ ε δ Dooli
1 − vf 1 + vf
(73)
The general equation (73) for the diffusivity depends on the loading type. The effect of the mechanical loading state on the diffusivity is summarized in Table 6. In Figure 18 is presented the effective diffusivity as a function of the parameter ψε for different mechanical loading expressed by the parameter δDooli.. We note that the starting points correspond to ψε=0 the value of composite materials with impermeable fibers. For tensile loading, expressed by δDooli >1, the difference between the effective diffusivities is not strongly significant. Table 6. effect of mechanical loading state on the diffusivity State
Free State Traction
σ=0
σ>0
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Compression
σ1
< ψ0
− per D εeff − per > D eff 0
ψ0
− per D εeff − per < D eff 0
− per D eff 0
Figure 18. Effective diffusion coefficient as a function of different stress states (vf = 0.7).
For compressive loading, expressed by δDooli 0 in fibre
0< r < rf t > 0 (76)
Initial, boundary, flux continuity, moisture content discontinuity conditions are given by:
⎧ m ∂Cm (rf , t) ∂Cf (rf , t) = Df ⎪Dε ∂r ∂r ⎪ m ⎪ f f ⎪C (r , t) = αε ρ Cm (rf , t) ⎨ ρf ⎪ ⎪Cm (rm , t) = Cm , ε ⎪ m f ⎪⎩C (r,0)= C (r,0)= 0
(77)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
where Cm and Cf are the the matrix and fiber moisture contents, Dm and Df are the matrix and fiber diffusivities. By putting ζ =
r r
m
, τ=
D εm r
m2
f t and φ = D , equation (77) becomes a dimensionless m
D
equation:
⎧ ∂C m ∂ 2 C m 1 ∂ C m + = ⎪ ζ ∂ζ ∂ζ 2 ⎪ ∂τ ⎨ f φ ⎡ ∂ 2 C f 1 ∂C f ⎪ ∂C = ⎢ 2 + ⎪ ∂τ δ ζ ∂ζ Dooli ⎣⎢ ∂ζ ⎩
in matrix ζ f < ζ < 1 t > 0 ⎤ ⎥ ⎦⎥
in fibre 0 < ζ < ζ f t > 0 (78)
And the initial, boundary, flux continuity, moisture content discontinuity conditions are rewritten as:
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
182
F. Jacquemin and S. Fréour
⎧ ∂Cm (ζ f , τ) φ ∂Cf (ζ f , τ) = ⎪ δDooli ∂ζ ∂ζ ⎪ ⎪ ⎪Cf (ζ f , τ) = α ρ m Cm (ζ f , τ) ε ⎨ ρf ⎪ ⎪Cm (1,τ) = Cm , ε ⎪ m f ⎪C (ξ,0) = C (ζ,0) = 0 ⎩
(79)
For impermeable fibers, φ must be taken equal to zero in the previous equations.
4.5. Moisture Content Estimation In this section, we attempt to estimate the moisture content in both cases of fibers. In case of permeable fibers, the parameters used in the moisture content evaluation are estimated by using (Rao et al., 1984) immersion data for epoxy/jute fiber composite in distilled water presented in Table 7. To determine the gap parameter α , the expression (58) is used and the moisture content of the fiber is determined from the expression (48) by using the data in Table 6. The value of α is equal to 3.8.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Table 7. diffusion data for epoxy/jute fiber composite immersed in distilled water Diffusion data (T=25°C) Moisture content (%) Diffusivity (mm2/s)
epoxy resin (LY556HT972) 3.2 8.3×10-8
epoxy/jute fiber composite (vf=0.7) 8.5 4.4×10-7
Figure 20. Average moisture content for various mechanical loads (impermeable fiber vf = 0.7).
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Mechanical States Induced by Moisture Diffusion …
For the parameter φ =
183
Df
, the value of the jute fiber diffusivity is needed which is Dm evaluated from the expression (59) and (60) for composite diffusivity in free state by using the data of Table 2 and the value of α determined from these data. The value of jute fiber diffusivity obtained is Df = 17.3161×10-7 mm2 /s, the diffusivity jute fiber value found in the literature is 17.7430×10-7 mm2/s (Aditya and Sinha, 1996) and the difference between the two values is about 2.4%. Finally φ = 21 .
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 21. Average moisture content for various mechanical loads (Permeable fiber vf = 0.7, φ = 21, α = 3.8 ).
In Figures 20 and 21 is plotted the average moisture content for different types of mechanical loading in the case of impermeable and permeable fibers. For both cases, we note an increase in moisture content at saturation for tensile loading witch is due to free volume increases. In other hand, a decrease takes place for compressive loading condition. As we note, the time to reach saturation is longer for permeable fiber than impermeable fiber, this is quite expected because in this case the diffusion occurred in both the matrix and fiber.
5. Conclusion and Perspectives In this work, for the first time, the evolution, as reported in the literature, of the macroscopic hygro-elastic properties of a composite ply, as a function of its moisture content, is taken into account in a scale-transition based approach dedicated to the prediction of the multiscale states of stresses experienced by the plies and the constituents of them, during the transient part of the hygroscopic loading of a composite structure. The scale-transition approach involves the inversion of the classical homogenisation procedure in order to estimate the evolution of the stiffness tensor of the epoxy matrix, as a function of its moisture content.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
184
F. Jacquemin and S. Fréour
The mechanical states predicted with the model accounting of moisture dependent properties were compared to the reference values obtained assuming the materials properties independent from the moisture content. The numerical computations show that, as expected, the longitudinal mechanical states (expressed in the reference frame of the ply) are unaffected by the fact of taking into account an evolution of the hygro-elastic materials properties as a function of the moisture content. This result is understandable, because the hygro-mechanical behaviour of carbon-fiber reinforced composite plies, is controlled by the reinforcements, in the longitudinal direction. Since the carbon fibers do not absorb water, their properties remain constant at any state of the moisture diffusion process. Thus, the longitudinal properties and mechanical states are independent from interactions between the moisture content and the hygro-elastic properties, in a fiber-reinforced composite structure. At the opposite, the estimated transverse and shear stress components, which strongly depend on the hygromechanical behaviour of the composite plies constituting matrix, the properties of which do vary as a function of the moisture content, can deviate from the reference values by up to 30%. As a conclusion, since the sizing of composite structures is strongly related to the amount of internal states of stresses predicted for the typical loads expected to occur during the service life, the present study demonstrates that the hygroscopic coupling relating the materials properties to the moisture content cannot be neglected, at least for composite structures designed for performing in humid environments. A multi-scale approach accounting for the existence of a hygro-mechanical coupling was used in the present work in order to achieve the determination of the time and space dependent internal stresses resulting from the purely hygroscopic loading of thin composite laminates. The considered coupling involves considering an evolution of the moisture transport process parameters as a function of the internal mechanical states, especially the volume strain of the organic matrix. The effective diffusion coefficient of the composite material is estimated from the homogenization procedure established by Hashin, accounting for the mechanical state dependent moisture diffusion coefficient of the constitutive epoxy. The hygro-mechanical coupling is assumed, as reported in the literature, to affect the maximum moisture absorption capacity, also. The in-depth moisture content evolution during the transient stage of the moisture diffusion process is deduced from Fick’s law. The macroscopic internal stresses are calculated from the classical continuum mechanics relationships whereas those of the plies very constituents (organic matrix and its reinforcements, respectively) are deduced from Eshelby-Kröner analytical self-consistent model. Accounting for the occurrence of the hygro-mechanical coupling yields a significant evolution of the moisture diffusion law governing parameters: the diffusion coefficient, and more noticeably the maximum moisture absorption capacity, each of them varying through the depth of the composite structure, also. The present study underlines that the choice of the geometrical stacking of the composite plies, since it enables to optimize the multi-scale mechanical states experienced by the composite structure, also enable to significantly optimize the parameters of the moisture diffusion process, so that the time and depth dependent moisture content profiles would be significantly weaker than the corresponding profile predicted in an unidirectional structure. Moreover, the hygro-mechanical coupled model leads to estimate multi-scale mechanical stresses being weaker than that predicted by the traditional uncoupled model. Thus, taking into account the existence of the organic matrix volume strain effects on the moisture diffusion process should be considered as a significant
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Mechanical States Induced by Moisture Diffusion …
185
parameter as regard to the sizing of composite structures conceived for withstanding hygromechanical loads during their service-life. As several experimental studies have shown (Neumann and Marom, 1985, Weitsman, 1987), the proposed modelling confirms that the presence of external mechanical loading have strong effect on the diffusion process in composites. By maintaining a Fick’s diffusion behavior, the theory of free volume leads to the theoretical interpretation of this behavior. A tensile loading involves an increase in moisture content at saturation and diffusivity and thus accelerate the process of diffusion. However, compression may cause a decrease in moisture content at saturation and in diffusivity and thereby slow the diffusion process. A more general form of the effective diffusivity expression function of the stress state has been established. This expression enables to determine, as a particular case, the effective diffusivity for free stress state. The next step concerning this axis of research will still deal with some additional physical factors in order to improve the realism and the reliability of the predictions obtained through the scale-transition models. For instance, the moisture diffusion process was assumed, in the present work, to follow the linear, classical, and established for a long time, Fickian model. Nevertheless, some valuable experimental results, already reported in (Gillat and Broutman, 1978), have shown that certain anomalies in the moisture sorption process, (i.e. discrepancies from the expected Fickian behaviour) could be explained from basic principles of irreversible thermodynamics, by a strong coupling between the moisture transport in polymers and the local stress state (Weitsman, 1990-a ; Weitsman, 1990-b). Thus, hygro-mechanical coupling satisfying the fundamental principle of thermodynamics will be investigated in the future. In further works, the evolution, reported in practice (Sai Ram and Sinha, 1991), of the organic matrix mechanical stiffness, as a function of its moisture content, will be considered in the multi-scale coupled approach based on the free-volume theory, presented in this work. Since according to (Youssef et al., 2009), this yields to the occurrence of an in-depth macroscopic properties evolution during the transient part of the diffusion process a dedicated, non iterative homogenization scale transition procedure based on Mori-Tanaka estimates (Fréour et al., 2006b) instead of the presently used Eshelby-Kröner model, would be required. Other published experiments reported that the measured diffusivities of carbon epoxy composites with long histories of exposure to sea water or to distilled water are higher by 3562 percent than the expected values. It was claimed that the considerable increase of the diffusivity had caused micro-damage in the composites, creating more channels of water penetration (Mazor et al., 1978). Moreover, fiber debonding (consequence of material damage) enhances moisture penetration by capillary flow along the interface, according to the observations achieved by Field and Ashbee (1972). As a consequence, a scale transition model accounting for damage occurring in such fiber-epoxy composites will be developed in further works, so that it would be possible to introduce damage related effects on the moisture diffusion behaviour of the material.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
186
F. Jacquemin and S. Fréour
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
References Aboudi, J. & Williams, T. O. (2000). A coupled micro-macromechanical analysis of hygrothermoelastic composites, International Journal of Solids and Structures, 37, 41494179. Abou Msallem, Y., Boyard, N., Jacquemin, F., Poitou, A., Delaunay, D. & Chatel, S. (2008). Identification of thermal and rheological properties of an aeronautic epoxy resinsimulation of residual stresses, International Journal of Material Forming, 1, 579-584. Adamson, M. J. (1980). Thermal expansion and swelling of cured epoxy resin used in graphite/epoxy composite materials, Journal of Materials Science, 15, 1736-1745. Aditya, P. K. & Sinha, P. K. (1996). Moisture diffusion in variously shaped fibre reinforced composites, Journal of Computers and Structures, 59, 157-166. Agbossou, A. & Pastor, J. (1997). Thermal stresses and thermal expansion coefficients of nlayered fiber-reinforced composites, Composites Science and Technology, 57, 249-260. Bao, L. R. & Yee, A. F. (2002). Moisture diffusion and hygrothermal aging in bismaleimide matrix carbon fiber composites-part I: uni-weave composites, Journal of Composite Science and Technology, 62, 2099-2110. Bowles, D. E., Post, D., Herakovich, C. T. & Tenney, D. R. (1981). Moiré Interferometry for Thermal Expansion of Composites, Experimental Mechanics, 21, 441-447. Chen, C., Han, B., Li, J., Shag, T., Zou, J. & Jiang, W. (2001). A new model on the diffusion of small molecules penetrants in dense polymer membranes, Journal of Membrane Science, 187, 109. Cohen, M. H. & Turnbull, D. (1959). Molecular Transport in Liquids and Glasses, Journal of Chemical Physics, 31(5), 1164-1169. Crank, J. (1953). A theoretical investigation of the influence of molecular relaxation and internal stress on diffusion in polymers, Journal of Polymer Science, 11, 151-168, 1953. Crank, J. & Park, G. S. (1968). in: Diffusion in Polymers. London : Academic Press Inc. Crank, J. (1975). The mathematics of diffusion, Clarendon Press, Oxford. Davies, P., Mazeas, F. & Casari, P. (2001). Sea water aging of glass reinforced composites. Journal of Composite materials, 35, 1343-1372. Derrien, K. & Gilormini, P. (2007). The effect of applied stresses on the equilibrium moisture content in polymers, Scripta Materialia, 56, 297-299. Derrien, K. & Gilormini, P. (2009). The effect of moisture-induced swelling on the absorption capacity of transversely isotropic elastic polymer-matrix composites. DiCarlo, J. A. (1986). Creep of chemically vapor deposited SiC fiber, Journal of Materials Science, 21, 217-224. Doolittle, A. K. (1951). Studies in Newtonian Flow. II. The Dependence of the Viscosity of Liquids on Free-Space, Journal of Applied Physics, 22, 1471-1475. Dyer, S. R. A., Lord, D., Hutchinson, I. J., Ward, I. M. & Duckett, R. A. (1992). Elastic anisotropy in unidirectional fibre reinforced composites, Journal of Physics D: Applied Physics, 25, 66-73. Eshelby, J. D. (1957). The Determination of the Elastic Field of an Ellipsoidal Inclusion, and Related Problems, In: Proceedings of the Royal Society London, A241, 376-396. Fahmy, A. A. & Hurt, J. C. (1980). Stress Dependence of Water Diffusion in Epoxy Resin, Polymer Composites, 1, 77-80.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Mechanical States Induced by Moisture Diffusion …
187
Ferreira, C., Casari, P., Bouzidi, R. & Jacquemin, F. (2006-a). Identification of Young Modulus Profile in PVC Foam Core thickness using speckle interferometry and Inverse Method, Proceedings of SPIE-The International Society for Optical Engineering. Ferreira, C., Jacquemin, F. & Casari, P. (2006-b). Measurement of the nonuniform thermal expansion coefficient of a PVC foam core by speckle interferometry - Influence on the mechanical behavior of sandwich structures, Journal of Cellular Plastics, 42(5), 393-404. Ferry, J. D. (1961). Viscoelastic properties of polymers, John Willey and Sons, Inc. Field, S. Y. & Ashbee, K. H. G. (1972). Weathering of fibre reinforced plastics. Progress of debonding detected in model systems by using fibres as light pipes, Polymer Engineering Science, 12, 30-33. Fréour, S., Gloaguen, D., François, M., Guillén, R., Girard, E. & Bouillo, J. (2002). Determination of the macroscopic elastic constants of a phase embedded in a multiphase polycrystal – application to the beta-phase of Ti17 titanium based alloy, Material Science Forum, 404-407, 723-728. Fréour, S., Gloaguen, D., François, M. & Guillén, R. (2003-a). Study of the Coefficients of Thermal Expansion of Phases Embedded in Multiphase Materials, Material Science Forum, 426–432, 2083–2088. Fréour, S., Gloaguen, D., François, M. & Guillén, R. (2003-b) Thermal properties of polycrystals - X-ray diffraction and scale transition modelling, Physica Status Solidi a, 201, 59-71. Fréour, S., Jacquemin, F. & Guillén, R. (2005-a). On an analytical self-consistent model for internal stress prediction in fiber-reinforced composites submitted to hygroelastic load, Journal of Reinforced Plastics and Composites, 24, 1365-1377. Fréour, S., Gloaguen, D., François, M., Perronnet, A. & Guillén, R. (2005-b). Estimation of Ti-17 βphase Single-Crystal Elasticity Constants using X-Ray Diffraction measurements and inverse scale transition modelling, Journal of Applied Crystallography, 38, 30-37. Fréour, S., Gloaguen, D., François, M. & Guillén, R. (2006). Application of inverse models and XRD analysis to the determination of Ti-17 βphase Coefficients of Thermal Expansion, Scripta Materialia, 54, 1475-1478. Fréour, S., Jacquemin, F. & Guillén, R. (2006-b). Extension of Mori-Tanaka approach to hygroelastic loading of fiber-reinforced composites - Comparison with Eshelby-Kröner self-consistent model, Journal of Reinforced Plastics and Composites, 25, 1039-1053. Fréour, S., Jacquemin, F. & Guillén, R. (2007). On the use of the geometric mean approximation in estimating the effective hygro-elastic behaviour of fiber-reinforced composites, Journal of Materials Science, 47, 7537-7543. Fujita, H. (1991). Notes on free volume theories, Polymer Journal, 23(12), 1499-1506. Gigliotti, M., Jacquemin, F., Molimard, J. & Vautrin, A. (2007). Transient and cyclical hygrothermoelastic stress in laminated composite plates: Modelling and experimental assessment, Mechanics of Materials, 39, 729-745. Gillat, O. & Broutman, L. J. (1978). Effect of External Stress on Moisture Diffusion and Degradation in a Graphite Reinforced Epoxy Laminate, ASTM STP, 658, 61-83. Gopalan, R., Rao, R. M. V. G. K. Murthy, M. V .V. & Dattaguru, B. (1986). Diffusion Studies on Advanced Fibre Hybrid Composites, Journal of Reinforced Plastics and Composites, 5, 51-61.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
188
F. Jacquemin and S. Fréour
Gueribiz, D., Jacquemin, F., Rahmani, M., Fréour, S., Guillén, R. & Loucif, K. (2009). Homogenization of moisture diffusive behaviour of composite materials with impermeable or permeable fibres – Application to porous composite materials, Journal of Composite Materials, 43, 1391-1408. Han, J., Bertram, A., Olschewski, J., Hermann, W. & Sockel, H. G. (1995). Identification of elastic constants of alloys with sheet and fibre textures based on resonance measurements and finite element analysis. Materials Science and Engineering, A191, 105-111. Hashin, Z. (1972). Theory of Fibre Reinforced Materials. NASA CR-1974. Herakovitch, C. T. (1998). Mechanics of Fibrous Composites, John Wiley and Sons Inc., New York. Hill, R. (1952). The elastic behaviour of a crystalline aggregate. Proc. Phys. Soc., 65, 349-354. Hill, R. (1965). Continuum micro-mechanics of elastoplastic polycrystals, Journal of the Mechanics and Physics of Solids, 13, 89-101. Hill, R. (1967). The essential structure of constitutive laws for metals composites and polycrystals, Journal of the Mechanics and Physics of Solids, 15, 79-95. Jacquemin, F. & Vautrin, A. (2002). A closed-form solution for the internal stresses in thick composite cylinders induced by cyclical environmental conditions, Composite Structures, 58, 1-9. Jacquemin, F., Fréour, S. & Guillén, R. (2005). A self-consistent approach for transient hygroscopic stresses and moisture expansion coefficients of fiber-reinforced composites, Journal of Reinforced Plastics and Composites, 24, 485-502. Jacquemin, F., Fréour, S. & Guillén, R. (2006). Analytical modeling of transient hygro-elastic stress concentration-Application to embedded optical fiber in a non-uniform transient strain field, Composites Science and Technology, 66, 397-406. Jedidi, J., Jacquemin, F. & Vautrin, A. (2005). Design of Accelerated Hygrothermal Cycles on Polymer Matrix Composites in the Case of a Supersonic Aircraft, Journal of Composite Structures, 68, 429-437. Jedidi, J., Jacquemin, F. & Vautrin, A. (2006). Accelerated hygrothermal cyclical tests for carbon/epoxy laminates, Composites Part A : Applied Science and Manufacturing, 37, 636-645. Kocks, U. F., Tome, C. N. & Wenk, H. R. (1998). Texture and Anisotropy, Cambridge University Press. Kröner, E. (1958). Berechnung der elastischen Konstanten des Vielkristalls aus des Konstanten des Einkristalls, Zeitschrift für Physik, 151, 504–518. Larobina, D., Lavorgna, M., Mensitieri, G. Musto, P. & Vautrin, A. (2007). Water Diffusion in Glassy Polymers and their Silica Hybrids : an Analysis of State of Water Molecules and of the Effect of Tensile Stress, Macromolecular Symposia, 247, 11-20. Loos, A. C. & Springer, G. S. (1981). Environmental Effects on Composite Materials, Moisture Absorption of Graphite – Epoxy Composition Immersed in Liquids and in Humid Air, 34-55, Technomic Publishing. Matthies, S. & Humbert, M. (1993). The realization of the concept of a geometric mean for calculating physical constants of polycrystalline materials, Physica Status Solidi b, 177, K47-K50. Matthies, S., Humbert, M. & Schuman, C. h. (1994). On the use of the geometric mean approximation in residual stress analysis, Physica Status Solidi b, 186, K41-K44.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Mechanical States Induced by Moisture Diffusion …
189
Matthies, S., Merkel, S., Wenk, H. R., Hemley, R. J. & Mao, H. (2001). Effects of texture on the determination of elasticity of polycrystalline ε-iron from diffraction measurements, Earth and Planetary Science Letters, 194, 201-212. Mazor, A., Broutman, L. J. & Eckstein, B. H. (1978). Effect of long-term water exposure on properties of carbon and graphite fiber reinforced epoxies, Polymer Engineering Science, 18(5), 341-349. Morawiec, A. (1989). Calculation of polycrystal elastic constants from single-crystal data, Physica Status Solidi b, 154, 535-541. Mori, T. & Tanaka, K. (1973). Average Stress in Matrix and Average Elastic Energy of Materials with Misfitting Inclusions, Acta Metallurgica, 21, 571-574. Morris, R. (1970). Elastic constants of polycrystals, International Journal of Engineering Science, 8, 49. Mura, T. (1982). Micromechanics of Defects in Solids, Martinus Nijhoff Publishers, The Hague, Netherlands. Neerfeld, H. (1942). Zur Spannungsberechnung aus röntgenographischen Dehnungsmessungen, Mitt. Kaiser-Wilhelm-Inst. Eisenforschung Düsseldorf, 24, 61-70. Neumann, S. & Marom, G. (1985). Stress Dependence of the Coefficient of Moisture Diffusion in Composite Materials, Polymer Composites, 6, 9-12. Neumann, S. & Marom, G. (1986). Free-volume dependent moisture diffusion under stress in composite materials, Journal of Materials Science, 21, 26-30. Neumann, S. & Marom, G. (1987). Prediction of Moisture Diffusion Parameters in Composite Materials Under Stress, Journal of Composite Materials, 21, 68-80. Patel, B. P., Ganapathi, M. & Makhecha, D. P. (2002). Hygrothermal Effects on the Structural behaviour of Thick Composite Laminates using Higher-Order Theory, Composite Structures, 56, 25-34. Rao, R. M. V. G. K., Chanda, M. & Balsubramanian, N. (1984). Factors Affecting Moisture Absorption in Polymer Composites Part II: Influence of External Factors, Journal of Reinforced plastic and composites, 3, 246-251. Reuss, A. (1929). Berechnung der Fliessgrenze von Mischkristallen auf Grund der Plastizitätsbedingung für Einkristalle, Zeitschrift für Angewandte und Mathematik und Mechanik, 9, 49-58. Sai Ram, K. S. & Sinha, P. K. (1991). Hygrothermal effects on the bending characteristics of laminated composite plates, Computational Structure, 40(4), 1009-1015. Shen, C. H. & Springer, G. S. (1977). Environmental Effects in the Elastic Moduli of Composite Materials, Journal of Composite Materials, 11, 250-264. Sims, G. D., Dean, G. D., Read, B. E. & Western B. C. (1997). Assessment of Damage in GRP Laminates by Stress Wave Emission and Dynamic Mechanical Measurements, Journal of Materials Science, 12(11), 2329-2342. Sih, G. C., Michopoulos, J. G. & Chou, S. C. (1986). Hygrothermoelasticity. Nijhoff, Dordrecht. Soden, P. D., Hinton, M. J. & Kaddour, A. S. (1998). Lamina properties lay-up configurations and loading conditions for a range of fiber-reinforced composite laminates, Composites Science and Technology, 58, 1011-1022. Tanaka, K. & Mori, T. (1970). The Hardening of Crystals by Non-deforming Particules and Fibers, Acta Metallurgica, 18, 931-941.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
190
F. Jacquemin and S. Fréour
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Tsai, C. L. & Daniel, I. M. (1993). Measurement of longitudinal shear modulus of single fibers by means of a torsional pendulum. 38th International SAMPE Symposium 1861-1868. Tsai, C. L. & Chiang, C. H. (2000). Characterization of the hygric behavior of single fibers. Composites Science and Technology, 60, 2725-2729. Tsai, S. W. & Hahn, H. T. (1980). Introduction to composite materials, Technomic Publishing Co., Inc., Lancaster, Pennsylvania. Tsai, S. W. (1987). Composite Design, 3rd edn, Think Composites. Vaughan, D. J. & McPherson, E. L. (1973). The effects of adverse environmental conditions on the resin-glass interface of epoxy composites, Composites, 4, 131. Voigt, W. (1928). Lehrbuch der Kristallphysik, Teubner, Leipzig/Berlin. Wang, Z. F., Wang, B., Yang, Y. R. & Hu, C. P. (2003). Correlations between gas permeation and free-volume hole properties of polyurethane membranes, European Polymer Journal, 39, 2345. Welzel, U., Fréour, S. & Mittemeijer, E. J. (2005). Direction-dependent elastic graininteraction models – a comparative study, Philosophical Magazine, 85, 2391-2414. Weitsman, Y. (1987). Stress assisted diffusion in elastic and viscoelastic materials, Journal of the Mechanics and Physics of Solids, 35, 73-93. Weitsman, Y. (1990-a). A Continuum Diffusion Model for Viscoelastic Materials, Journal of Physical Chemistry, 94, 961-968. Weitsman, Y. (1990-b). Moisture in Composites: Sorption and Damage, in: Fatigue of Composite Materials. Elsevier Science Publisher, K.L. Reifsnider (editor), 385-429. Williams, M. L., Landel, R. F. & Ferry, J. D. (1955). The temperature dependence of relaxation mechanisms in amorphous polymers and other glass-forming liquids, Journal of the American Chemical Society, 77, 3701-3707. Youssef, G., Fréour, S. & Jacquemin, F. (2009). Effects of moisture dependent constituents properties on the hygroscopic stresses experienced by composite structures, Mechanics of Composite Materials, 45, 369-380.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
In: Composite Materials in Engineering Structures Editor: Jennifer M. Davis, pp. 191-228
ISBN: 978-1-61728-857-9 © 2011 Nova Science Publishers, Inc.
Chapter 4
FATIGUE AND FRACTURE OF SHORT FIBRE COMPOSITES EXPOSED TO EXTREME TEMPERATURES B.G. Prusty and J. Sul School of Mechanical and Manufacturing Engineering, University of New South Wales, Sydney, NSW, Australia
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Abstract Fibre-reinforced composites have been used for more than 50 years and are still being evolving in terms of material integrity, manufacturing process and its performance under adverse conditions. The advent of graphite fibres from polyacrylonitrile organic polymer has resulted in a high performance material, namely carbon based composites, performing better in every respect than glass fibre-reinforced plastic (GFRP). However, glass fibres are still in high demand for wide applications, where the cost takes precedence over performance. Owing to its quasi-isotropic properties, randomly orientated short fibre reinforced composites, particularly chopped strand mat (CSM) and sheet moulding compound, are playing a critical role in boat building industry and automotive industry, respectively. As structural performance of composite material is being improved, GFRPs are expected to replace metals in more harsh applications, in which high cyclic loadings and elevated temperatures are applied. Furthermore, heat deflection temperature of common thermosetting resin is in the range from 65ºC to 85ºC under applied stress of 1.8MPa. The thermal effects on short-fibre thermosetting composites have not been flourishingly investigated. Fatigue prediction of mechanical structure is not only critical at the design stage, but is much more critical for the maintenance strategy. The fatigue, fracture and durability of GFRP-CSM are complex issues because of so many variables contributing to thermal and mechanical damages. Despite a number of approaches to modeling fatigue damage of GFRP using phenomenological methodologies based on the strength and stiffness degradation, or physical modelling based on micro-mechanics, their performance under adverse thermo-mechanical loading has not been fully understood to benefit the end users.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
192
B.G. Prusty and J. Sul
1. Introduction
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Fibre-reinforced composites are the materials of choice in many engineering structures experiencing repetitive loading in their life time, such as the airplane fuselage, boat hulls and even in building structures. Fatigue behaviour of composite materials has not been a major issue in the past due to the low working strains. However, since fibrous composites are the most promising materials to replace conventional materials whose specific strength and stiffness are relatively low, a number of engineers and investigators have raised a question with regard to fatigue performance of composites whilst the knowledge achieved is not sufficiently comprehensive. This is because composite materials are inhomogeneous and anisotropic unlike traditional materials, such as steel and alloys, of which fatigue progresses from the initiation of a single crack. Fatigue refers to the effect of cyclic or intermittently alternate stresses. Cyclic stress due to either repetitive mechanical loads or due to alternate heating and cooling, or even to both mechanical and thermal loadings, is evidently more adverse to fibre-reinforced composites than monotonic loading. Under cyclic loadings, damages are accumulated in a general fashion and cracks induced by fatigue are initiated at localised sites within the component. These cracks and damages do not always occur by the propagation of a single macroscopic crack. Eventually, they expand in size and coalesce to such an extent that the residual constituents are not able to support the stresses. Therefore, highly fatigued fibre-reinforced composite components suddenly fracture in most cases as a result of microscopic damages. The nature of fatigue failure of composites is not always distinctive as that of homogeneous materials. Composite materials instead are considered to be failed when composite has lost its elastic modulus by 70% compared to the initial value. Fatigue loading brings about micro-cracks in polymer matrix mainly contributing to loss in stiffness due to • • • •
severance of polymer chain as a consequence of intense localised stress, the accumulation of heating due to hysteresis, the re-crystallisation of material as a result of extensive movement of chain structure and accumulative crack generation or multiple crack formation.
Hysteresis is of particular importance during crack propagation in thermosetting polymers when a crack moves through a body element close to the crack tip will undergo a full deformation cycle. Therefore, it is important to note that fatigue fracture caused by an alternating stress with amplitude is significantly lower than that required for tensile static fracture. The prediction and description of fatigue behaviour have been restricted to continuous fibre composites albeit the application of short fibre-reinforced composites becomes more diverse, in which the structure is subjected to multi-axial loadings and cost effectiveness is mainly required because it is more amenable to mass-production techniques than continuous fibre composites. However, it is difficult to provide a complete account of the stress-strain response and the final fracture strength of randomly oriented short fibre reinforced composites. This chapter deals with fatigue and fracture mechanisms of fibre-reinforced
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
193
composites, the effect of elevated temperature on composites, characteristics of short-fibre composites, damage modelling of short-fibre composites and its verification.
2. Fatigue and Fracture of Composite
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Polymer matrix composite can have a number of fibre orientations, namely continuous, discontinuous or randomly orientated. The orientation of fibre which comprises the composite together with matrix plays a significant role in the failure modes that composite structures may experience. For unidirectional continuous fibre composites, the direction of externally applied stress largely determines the failure mode by which the composite will fracture. However, in laminates comprised of a number of plies with varying orientations, the failure mode is the result of a complex interaction of factors. These interactions can give rise to matrix cracking, delamination, fibre fracture, de-bonding, matrix crazing, void growth, multidirectional cracking, etc. Figure 2.1 shows an example of composite failure by fibrematrix de-bonding. The occurrence of damage regions which are considered as discontinuities in a composite can in fact advance toughness of the composite. This is because the internal discontinuities absorb energy and lead to a redistribution and relaxation of the applied stress. Fibrereinforced composites can contain a wide range of such discontinuities, most of which form during loading. For instance, generally detected discontinuities include fibre breakages and micro-cracks within the matrix. Numerous types of damage modes in composites have been identified as important energy-absorbing mechanisms including fibre de-bonding, fibre pullout, delamination, fibre breakage and matrix cracking. Of these, delamination and fibre pullout make up the major mechanisms of energy dispersion during failure of a composite. Furthermore, energy absorption in short-fibre composites takes place by the processes of both yielding of the matrix and fibre pull-out.
Figure 2.1. Electron microscopy images of Fibre-matrix de-bonding [1].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
194
B.G. Prusty and J. Sul
2.1. Failure Mode of Composites The failure behaviour of fibrous composites is complicated and can be varying with the constitution of the matrix and fibre, the fibre volume fraction, the nature of the interfacial bond, fibre orientation, stacking angle and sequence, level of void content and type of loading, as well as the chemical conditions. Scheirs [2] clearly classified each common failure mode of fibre-reinforced composite as below, including de-bonding, matrix and fibre cracking, interfacial-bond failure and delamination.
De-bonding De-bonding takes place due to interfacial failure along the fibre-matrix boundary. This is characterised by the fracture surface showing numerous protruding fibres with little or no resin adhering to them, as well as the presence of smooth channels in the matrix.
Interlaminar Failure Interlaminar failure can occur when the interfacial strength between matrix and fibres is greater than the matrix cohesive strength (Figure 2.2). Such failure is manifested in the composite exhibiting excessively brittle behaviour. Interlaminar splitting of composites is more prone to occur when there is a high void content. Something as seemingly innocuous as a spanner being dropped on a composite sheet can initiate undetected damage, which may result in failure by buckling, when the sheet is loaded in compression. Longitudinal splitting can occur in composites reinforced with unidirectional glass fibres. Interestingly, this mode of cracking does not always result in catastrophic failure.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Fibre Buckling Fibre buckling can occur in compression when the matrix has inadequate strength. For this reason, a strong matrix is desirable as well as one with a high glass transition temperature ( Tg ), so that the composite will have good compressive properties at elevated temperatures.
Figure 2.2. Interlaminar fracture of Carbon/Epoxy composites using Scanning Electron Microscopy [3]. Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
195
Micro-buckling of fibres is a common form of failure in case of continuous fibrereinforced composites subjected to compression. The fibre undergoes deformation in a sinusoidal fashion under compressive loads when supported by a surrounding matrix. Fibre micro-buckling has also been found to occur as a result of curing at high temperatures when there is a significant difference between the degree of contraction of the matrix and the fibres.
Fibre Pull-out Fibre pull-out arises due to variations in the interfacial bond strength and localised load transfer from the fibre to the matrix. The contribution of fibre pull-out to the overall failure of the composite can be ascertained by SEM examination of its fracture surface. Fibres that appear clean and leave smooth channels in the matrix are those fibres which exhibit poor adhesion, while fibres with adherent matrix debris represent those fibres which possess considerable adhesion to the matrix. The energy dissipated during fibre pull-out from the matrix is largely dependent on the degree of interfacial friction present, with this in turn being determined by the shrinkage forces which arise on cooling of the composite.
Fibre Breakage Fibre breakage has been regarded as a significant energy-absorbing mechanism in some composites. Unfortunately, glass fibres have a high modulus and are brittle in nature, thus their contribution to overall energy absorption is limited.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Cracking of Composites For a crack to grow, the energy released in the matrix during each increment of crack growth should be at least as much as the energy consumed in creating the new crack surfaces. Cracking in composites can be initiated by de-bonding at the fibre-resin interface. This can result in a transverse ply crack which increases in length, and upon reaching a fibre continues along it and then proceeds back into the matrix. Since the crack length has deviated from its path a number of times, and travelled a considerable distance, there is a progressive reduction in the modulus of the composite. This, however, does not necessarily make the material much weaker than before. As a result of such extensive internal cracking, the composite can turn milky white, due to the reflection of light from the surfaces of these internal cracks. This phenomenon, where a propagating crack in the matrix can be deflected a number of times when it impinges on the fibre reinforcement, known as crack deflection. As a result, the crack can be deflected to a considerable distance from the initial plane of fracture.
Micro-cracking of Composites The micro-cracking of continuous fibre composites is quite a common problem. This is due to the strains induced by thermal expansion mismatches between the fibre and the matrix. This phase occurs during the cooling phase of the composite from its curing temperature and is especially important in high-temperature composites. The thermal expansion mismatch between the matrix and the fibre can cause considerable stresses (known as thermal strains), and these can result in complete yielding of the matrix, and hence micro-cracking.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
196
B.G. Prusty and J. Sul
Mechanical properties of composites vary as micro-cracks occur. Furthermore, it is important to note that macroscopic thermal mismatches can also occur between cross-plies. Cross-ply laminates generate higher residual stresses than unidirectional laminates because of the extensive anisotropy in the thermal expansion of the plies. For example, the 0º plies impose constraints on the 90º plies, thus causing micro-cracking. Cure-induced micro-cracking has been observed in glass-resin composites, graphite-epoxy composites and carbon-polyimide composites. The type of packing and the fibre spacing are important factors to determine the magnitude of residual thermal stresses in a composite system.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
2.2. Fatigue Failure of Composites Fatigue failure in polymer composites is characterised by means of a progressive loss of stiffness. It is obvious that this behaviour is quite different to the effect of fatigue on metals in which fatigue damages are accumulated in a localised fashion. In general, the fibres are arranged in parallel to the direction of the highest stresses anticipated, or also common for the fibres to be oriented randomly within the thermosetting polymer matrix. The former materials are known as anisotropic materials, for which directional properties vary, and adjusting the anisotropy is a key manner to control the material properties for specific applications. In other words, the properties of fibre-reinforced composites are determined not just by the properties of each fibre and matrix, but the orientation and distribution of the fibres. The matrix plays essential roles in transferring loads from fibre to fibre as well as in binding the fibres together. Nonetherless, the composites of which matrix is relatively stiff has worse fatigue resistance than that of flexible matrix composites. An ordinary feature of fatigue failure is due to diffusion which arises early in the damage development of composites. In case of homogeneous materials, the accumulation of cracks led by fatigue propagates in the perpendicular direction to the direction of applied stress. On the other hand, in case of heterogeneous composite materials, fatigue loading results in various types of failure modes, detailed in the previous section, which lead to a widely scattered damage area. There is a notable difference between conventional materials and fibre-reinforced composites in terms of fatigue behaviour that in the former the extent of damage grows statically and constantly while in the latter crack propagation advances with a progressive decrease in the stress of the composite and the dispersed damages. Together with fibre breakage and matrix cracking, a fatigue failure mode that appears in composites under cyclic loading in common is de-bonding of fibres especially those oriented orthogonally to the cyclic loading direction. A number of micro-mechanisms cause micro-cracks as well as crazes that are accumulated with each fatigue cycle, which define fatigue failure of composites. Fatigue behaviour is of particular importance for short-fibre-reinforced polymer composites. The structural components in which an isotropic material is required are becoming ever increasingly widespread. This is because uniaxial fibre-reinforced composites only show outstanding fatigue resistance in the circumstances where stresses are applied in the parallel direction to the fibre axis. Therefore, the application of unidirectional composites in engineering structures are somehow limited, e.g. helicopter and turbine blades, and so forth. Not only should mechanical stresses be considered in the fatigue study, but the thermal stresses also play a critical role in accumulating fatigue damage on the fibre-reinforced composites. Cyclic loading of composites can cause internal heat build-up. This is not easily
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
197
dissipated because of the low thermal conductivity of polymer composites. The degree of heat build-up is dependent on the frequency of the cycles and could exceed 20Hz. The higher thermal conductivity of carbon fibres is one of the reasons why carbon-fibre-reinforced composites have superior fatigue resistance to that of glass-filled composites [2]. This is explained by the empirical fact that at a stress amplitude of 300 MPa, 3:1 (glass : carbon) hybrid composites have a fatigue life of about 100 times that of all-glass composites [4].
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
3. Short Fibre Composites The flourishing demand for continuous fibre reinforced composites in aerospace applications, in which high performance materials are particularly required, has drawn close attention from many investigators to the fatigue behaviour. In contrast, glass short fibre composites in particular have evolved through a more assorted range of applications without definite confidence in fatigue performance. Until the 1980s, only a few investigators were interested in elemental research on the fatigue behaviour and performance of short fibre composites. However, faithful efforts of engineers to reduce the weight of engineering structure, especially in automotive industry, have led to ever increasing application of shortfibre reinforced plastics which have promoted considerable enthusiasm of many investigators in SMC (sheet moulding compounds) for body parts with short glass fibre based and reinforced injection-moulded thermoplastics for engine parts. In addition, the common fabrication methods, such as vacuum bag, hand lay-up, filament winding, etc, are generally specialised for unidirectional composites, which are generally not suitable for mass production, but for short runs or custom-built products demanding high production cost. Along with a cost problem, for large number of articles with complex shapes, foregoing fabrication techniques are appropriate, so that injection, compression and transfer moulding were developed. In consequence, the price to be paid for the use of such mass production techniques is a shortening of fibre length [5]. The reduction in fibre length is partly due to the requirements of the processing technique, but some processes which involve mechanical shearing and mixing actions also promote considerable fibre breakage. Fibre damage is particularly noticeable for injection moulding, extrusion and mixing of polyester moulding compounds. Led by automotive applications, penetration of short fibre composites into fatigue sensitive applications has steadily increased in a variety of industries for several reasons as they: (1) can be moulded into complex shaped parts (for which continuous fibre composite fabrication is impractical) with improved performance and/or economics when compared with metals or unreinforced thermoplastics. (2) can be processed at the high production rates required for automotive applications. (3) can have planar isotropic properties which are competitive with planar quasiisotropic continuous fibre systems. (4) are available with a variety of high performance thermoplastic matrices developed in recent years, which can provide a broad range of mechanical, thermal and environmental properties. The fatigue resistance of the most short fibre composites is similar to that of continuous glass or carbon fibre composites in the off-axis directions. Short fibre composites have some Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
198
B.G. Prusty and J. Sul
critical shortcomings compared to continuous fibre composites. Their modulus and strength are inherently lower than those of continuous fibre systems in the fibre direction due to the presence of fibre ends in the matrix, so that matrix is more likely to be stressed in order to carry the loads from fibre to fibre. Moreover, the short fibre composites are more sensitive to notch in any shape than continuous fibre composites, depending on the type of fibre and matrix. Chopped strand fibre composites have a problem with inconsistent properties along with notch sensitivity and poor controllability of fibre length and orientation. Fatigue damage can be modelled using fracture mechanics and fatigue crack growth theories form homogeneous materials since short fibre composites tend to fail by the development and propagation of a single macroscopic crack in contrast to continuous fibre reinforced composites.
3.1. Fibre Length and Orientation
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
With the exception of continuous reinforcement, the lengths of all fibre reinforcements are not exactly uniform. Due to this inconsistency, the term ‘short fibre’ should be clarified so as to differentiate it with ‘long fibre’. The fibre length, hence, should be considered in correlation to the material parameter that is known as ‘the critical fibre length’. Matthew et al. [6] defined that the critical fibre length is a function of the matrix and the reinforcement and as such varies considerably for different composites. It is therefore possible for fibres of 5mm length to be classified as short in one system and not in another. The behaviour of short fibres in general is dominated by end effects and they do not therefore act as good reinforcing agents. Discontinuous fibres are normally supplied by manufacturers in standard lengths for different processing routes. Given that a typical fibre may have a diameter of approximately 10 μ m , it is clear that high levels of length degradation are required to reduce them to ‘particles’. This is because processing techniques such as injection moulding have a devastating effect on fibre length. It is clear that dependent upon the type of material used, and the method chosen to process it to its final shape, a wide variety of fibre lengths will be present. Whilst fibres even down to 50 μ m in length may retain some ability to reinforce, it is the fact that actual fibre length and its distribution are uncertain that can cause design problems. Fibre orientation and distribution are just as important as the length of fibre. In spite of the common misunderstanding that fibre orientation effects in short fibre system is insignificant compared with unidirectional composites, they should be taken into consideration. The fabrication process is the key determinant of the fibre orientation. Stiffness and strength of the laminate comprising of unidirectional continuous fibre composites are comparatively predictable using micromechanical modelling. In case of randomly oriented short fibre composite laminates, due to their nature and the fabrication process, their properties are varying in the perpendicular direction to the plane of the lamina showing anisotropic characteristic. On the other hand, according to Matthew et al. [6], composites whose fibres are short, and processing methods involve flow of material in a mould, change in fibre orientation throughout the moulding is inevitable. This applies to Bulk Moulding Compound (BMC) and a wide range of reinforced thermoplastics. The orientation of the fibres may be impossible to predict for the composites in which thick or variable sections and several injection points are involved. In any case the properties of the material could differ markedly from area to area within the same moulding. Changes in fibre orientation are related to a number of factors, such as the geometrical properties of the fibres, viscoelastic behaviour
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
199
of the fibre-filled matrix, mould design and the change in shape of the material produced by the processing operation. In many processing operations the polymer melt, or charge, undergoes both elongational (or extensional) flow and shear flow.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
3.2. Stress and Strain Distribution at Fibre Damage mechanisms which take place during the life of composite laminates change the local geometry of the laminates in the structure. The continuity of the materials is interfered by the damage on the individual fibre. In the studies of the elastic properties of unidirectional continuous composites, the effects of fibre ends in the matrix have been disregarded because the fibres only end at the surface of the composites. However, owing to the decrease in the aspect ratio (ratio of fibre length l to the fibre diameter) of fibres, the fibre end effects play a critical role in deteriorating the efficiency in stiffening and reinforcing the polymer matrix. This is because the matrix enclosing fibres is varied by the discontinuity. Many previous investigators have been neglecting the effects of fibre ends of continuous fibre composites. Nonetheless it may be partly responsible for the fracture study of continuous fibre composites because fibre ends may exist once continuous fibres break down into discontinuous portions. Cox [7] considered a fibre of certain length embedded in a matrix of lower modulus as depicted in Figure 3.1. It is assumed that the fibre is bonded properly with the matrix and an applied stress on the resin is transferred to the fibre at the interface. The matrix and the fibre will experience different tensile strains due to their difference in moduli. In other words, the strain in the fibre in the region of the fibre ends will be less than that in the matrix, as illustrated in Figure 3.2. Shear stresses are induced around the fibres in the direction of the fibre axis and the fibre is stressed in tension as a consequence of this difference in strain between the fibres and the matrix. The shear strength of the fibre-matrix interface is relatively low, typically of the order of 20 MPa, although it can exceed 50 MPa, for a polymer matrix composite. However the surface area of the fibre is large, so that, given sufficient length, the fibre can carry a significant load, even up to the fibre fracture load [6]. Cox [7] has also shown analytically that the stress distribution along a fibre aligned parallel to the loading direction of the matrix can be represented in Figure 3.2. The assumptions made in this analysis are that the fibre and matrix only deform elastically and the interface is thin and gives good bonding between the fibre and the matrix. According to the reproduction in Figure 3.2 based on the analysis proposed by Cox [7], the tensile stress is zero at fibre ends, and for a sufficiently long fibre falls almost zero in the centre. It is variation of shear stress (‘shear effect’) that causes the build-up of tensile stress in the fibre. Meanwhile, Hull [8] stated that the shear stress is a maximum at the fibre ends and falls almost to zero at the centre. These results show that there are regions at the ends of the fibre which do not carry the full load so that the average stress in a fibre is less than that in a continuous fibre subjected to the same external loading conditions. The reinforcing efficiency decreases as the average fibre length, l , decreases because a greater proportion of the total fibre length is not fully loaded. The maximum possible value of strain in the fibre is the strain, ε , applied to the composite material as a whole so that the maximum
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
200
B.G. Prusty and J. Sul
stress in the fibre is strain times elastic modulus of fibre, namely ε E f . To achieve this maximum stress the fibre length must be grater than a critical length, lc . The critical fibre length may be defined as the minimum fibre length for a given diameter which will allow tensile failure of the fibre rather than shear failure of the interface, i.e., the minimum length of fibre required for the stress to reach the fracture stress of the fibre. The schematic representation in Figure 3.2 shows that fibres longer than lc in the regions at the ends of the fibre which are not fully loaded have a length lc / 2 .
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 3.1. Diagrammatic representation of deformation around a discontinuous fibre embedded in a matrix subjected to a tensile load parallel to the fibre [8].
Figure 3.2. Variation of tensile stress in a fibre and shear stress at the interface (The figure was reproduced on the basis of the theory by Cox [7]).
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
201
The theoretical analysis and the empirical data prove that fibre end regions do not carry the full load unlike the centre region of the fibre. This is due to the fact that the average stress in short fibre is distinctively less than that in continuous fibres subjected to the same mechanical loadings. Continuous fibre, hence, is much more efficient to endure external loads only in case that the mechanical load is in the same direction to the fibres. The interface strength, meaning robust interfacial bonding between fibre and matrix, is of importance to the reinforcing efficiency of fibres along with the superior profile of fibre itself. The reduction in tensile stress at the fibre end results in the large shear stresses which may lead to unfavourable consequences, including shear yielding, debonding at the interfaces, and cohesive failure of matrix and fibre.
4. Mechanical Property Variation in Fatigue of Polymer Based Short Fibre Composites
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Without the fatigue mechanisms of continuous fibre system, it is impossible to build up the theory in regard to the fatigue process and behaviour of short-fibre composites. Fatigue behaviour of continuous fibre composites have been researched sufficiently more than that of short fibre composites under fatigue loading. Of particular importance is the chopped strand nature of the CSM-SMC group. The fibres in a strand are often tightly bound together in a bundle, so that the strand acts as a single large diameter fibre. This phenomenon alters the aspect ratio of the short fibre strands, according to Mandell [9], from the range of 1000 times more for their individual fibres to an effective range of 10-30, and the behaviour becomes dominated in most cases by the matrix and interface. Low effective length to diameter ratio further limits the range of available systems studied in fatigue.
4.1. Residual Strength of Short Fibre Composites The cyclic fatigue response of engineering materials has been a topic of interest to many investigators. Cyclic or monotonic loadings cause damage on the materials that leads to degradation in strength. The remaining strength after the reduction in strength is known as ‘residual strength’ which is considered to be the one of the everlasting approach for the most conventional materials and it is of the most effective and simple method to predict the final failure stage of material. Although defining strength in fibrous composite materials cannot generally be done by simply identifying a single ‘stress level’ that causes failure [10], this approach is similarly applied to the fibre-reinforced composite materials. Later in life the amount of damage accumulated in some region of the composite may be so great that the residual load-bearing capacity of the composite in that region falls to the level of the maximum stress in the fatigue cycles and failure occurs once the residual strength coincides with cyclic stress, as shown in Figure 4.1. This process may occur gradually, when it is simply referred to as degradation, or catastrophically, when it is termed ‘sudden-death’. Changes of this kind do not necessarily relate to the propagation of a single crack, and this must be recognised when attempting to interpret composites fatigue data obtained by methods developed for metallic materials.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
202
B.G. Prusty and J. Sul
Figure 4.1. Degradation of composite strength by wear-out until the residual strength σ R falls from the normal composite strength σ C to the level of the fatigue stress, at the point where failure occurs [11].
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
In case of a micro-crack existing in highly anisotropic composites, it may or may not propagate under the action of a cyclic load, depending upon the nature of the composite. The crack will often refuse to propagate normal to the fibres (mode 1) but will be diverted into a splitting mode, sometimes resulting in end-to-end splitting which simply eliminates the crack. In contrast with continuous fibre composites, in GRP laminates containing woven-roving or chopped-strand mat reinforcement, crack tip damage may remain localised by the complex geometry of the fibre array and the crack may proceed through this damaged zone in a fashion analogous to the propagation of a crack in a plastically deformable metal [11]. The fatigue behaviour of short-fibre composite is dominated by complex stress distributions due to a discontinuous property of fibre. The stress concentration at the region of fibre ends often deteriorates the strength of short fibre reinforced composite compared to the continuous fibre composite having the same fibre / matrix combination and fibre volume fraction.
4.2. Elastic Modulus of Short Fibre Composite Materials Under either low levels of monotonic stress or low cyclic stress, literally most composites regardless of the type of fibre and matrix experience damage distributed in the stressed region. Although the residual strength provides a relatively accurate estimation for the remaining life of the composite structures, strength is not always immediately reduced as the damage is dispersed. On the other hand, the dissipated damages can often lead to the instant reduction in elastic modulus even at the low cycles of loading, which was experimentally proved and shown in Figure 4.2; even at the low cycle fatigue of 5000 cycles the elastic modulus in the last 5 cycles of the total is notably lower than the initial modulus due to the permanent deformation. In addition, the stiffness of materials is measurable without destruction of engineering structures. These facts can make the elastic modulus more advantageous than strength as a damage indicator, since low-load long-life specimens or components undergo greater damage than high-load short life specimens for the same fraction of life [12]. The low-load condition is more common for applications whereas the high load condition is more common for laboratory characterisation. Changes in stiffness of composites caused by the extensive matrix cracking can be substantial, tens of percent depending on the
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
203
details of the material [12]. The monotonic decrease in stiffness during the life of composites is not accompanied by a monotonic decrease in strength [13]. Point-wise stiffness may be a function of time or cycles of applied loading at the local level, i.e. stiffness changes due to processes like micro-cracking and creep are likely to influence the load-direction normal stress in the zero degree plies which control the remaining strength of the notched laminate. Unlike randomly oriented short fibre composite, matrix cracking in the 90º plies of a cross-ply laminate will ‘shed’ normal stress onto the 0º plies, but matrix cracking and delamination near a notch can relax the local stress concentration. Hence, reduction of the stiffness of the 90º plies as a function of cycles of loading is an experimental characterisation that must be entered into an iterative analysis of the stresses in the 0º plies, as a function of cycles of loading. Stiffness changes during cyclic loading typically have the form shown in Figure 4.3 [12] which is generally explained as three degradation stages, viz. (a) a dramatic reduction stage, (b) a stable and gradual reduction stage and (c) a failure stage [14]. In addition to the reduction in modulus of elasticity due to micro-cracks in matrix, creep caused by viscoelastic properties of matrix-dominated composites also deteriorates the stiffness of the laminate. As a result, changes in stiffness are not sufficient to predict life [12]. Other characterisations are needed if there are other processes that contribute to a change of stress state or material state. Nevertheless, variations in stiffness are an indispensable part of damage parameter to estimate the remaining life. Stiffness as a function of position does determine local stress and strain distributions for a given loading, so tracking and modelling the large stiffness changes that can occur in composite materials for acceptable service conditions is a critical part of a viable life prediction model. First 5 cycles Last 5 cycles
25
15
10
Stress (MPa)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
20
5
0 0.0027
‐0.0012
0.0010
0.0012
‐0.0014
0.0027
‐0.0012
0.0010
0.0013
‐0.0014
0.0027
‐0.0012
0.0010
‐5
‐10
‐15
Strain (mm/mm) ‐20
Figure 4.2. Behaviour of short E-glass fibre / polyester under constant cyclic loading up to 5000 cycles [14].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
204
B.G. Prusty and J. Sul
Figure 4.3. Typical stiffness change of 90º plies in a laminate under cyclic loading [12].
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
4.2.1. Elastic Properties of Short Fibre Composite Materials Using Rule-of-Mixtures The Rule-of-mixtures is a simplified mathematical approach to estimate an upper limit of elastic modulus of the composites by considering the elastic moduli of individual reinforcement and matrix. A simple rule of mixtures equation was suggested by Hull [8]. Since the reinforcing efficiency of short fibres is less than that of long fibres, the effective modulus of short fibre composite materials will be adversely affected. In general, a material has a three-dimensional distribution of fibre orientations and a distribution of fibre lengths. There is no satisfactory description of the elastic properties in terms of these parameters. On the basis of the rule-of-mixtures for the unidirectional composite materials, the rule-ofmixtures equation for short fibre composite is expressed. For a unidirectionally aligned material containing fibres of length l , the rule of mixtures equation may be modified with the inclusion of a length correction factor, ηl , so that E11 = η l E f V f + E m (1 − V f )
(4.1)
For randomly oriented fibre systems there is a distribution of fibre orientation and the reader will not be surprised to learn that the reinforcing efficiency of the fibres is reduced further. The length correction factor, ηl , can be negligible for short fibre composite as it is virtually unity in most case where the length of individual fibre is longer than 1 mm. Krenchel [5] introduced an orientation efficiency factor , η0 , into equation 4.1 to account for the further reduction in efficiency, Ec = η 0η l E f V f + E m (1 − V f )
(4.2)
Values of η0 have been reported in table 4.1 for simple fibre orientation distributions assuming elastic deformation of the matrix and fibres, and equality of strains. These values presented in table 4.1 show that the contribution from the fibres is reduced by almost a half when the orientation is changed from random-in-plane to three-dimensional random.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
205
Table 4.1. Orientation efficiency factor η0 for several systems [5] Orientation of fibres Unidirectional in longitudinal direction Unidirectional in transverse direction Two-dimensional random in plane Three-dimensional random
η0 1 0 0.375 0.2
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
5. Temperature Effect on the Thermosetting Polymer Failure in thermosetting polymers can occur due to neglect of a temperature effect in fatigue. The majority of reinforcement used for high performance engineering structures is usually infused into a thermoplastic with a comparably high fibre volume fraction. In other words, the thermoplastic composites are designed to be fibre dominated as opposed to thermosetting composites using glass fibre as reinforcement less dominating the fibre fraction, which is in turn strongly influenced by the fact that thermosetting polymer is commonly sensitive to the variation of temperature whilst glass fibre reinforcement is relatively insensitive to temperature. This is clearly supported by Weeton [15] arguing that Eglass fibre, for instance, has good strength, stiffness, electrical and weathering properties as well as E-glass retains its properties up to 250 ºC, while mechanical properties of polyester matrix deteriorates above 75 ºC. Experimental results conducted by Reifsnider and Pastor [16] provide the evidence that the tensile strength in the fibre direction of a polymer composite coupon can change by 15-34% when the matrix properties or the fibre/matrix coupling changes as a result of temperature or local constituent variations, even though the fibres are unaffected by those changes. They also stated that the researchers must take those failure modes into consideration, which are mainly caused by applied environments such as temperature, chemical agents, and time or cycles. The failure mode must be determined for the conditions to be modelled by the experimental characterisation.
5.1. Thermosetting Polymer Thermosetting polymers are appeared to be the most applicable polymers to glass fibres amongst commonly used polymers. Densely cross-linked thermosets are usually used below their glass transition temperatures. Strong bonds of the cross-links haul the polymer chains together which restrains the chains from the movement. Therefore, it is well known that thermosets are very brittle and intractable materials at the temperature below their glass transition temperatures. The most widespread thermosetting polymers are epoxy, unsaturated polyester, phenol-formaldehyde, and vinylester for marine applications due to its exceptional resistance to water. In spite of diversity of thermosets, they have a number of characteristics in common. Matthews and Rawlings [17] described common thermosetting polymers in detail, starting with phenolics which represent about 43% of thermoset market [18].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
206
B.G. Prusty and J. Sul
Phenolic Resins They are the oldest of the thermosets discussed but, nevertheless, due to their low cost and good balance of properties together with their good fire resistance, they are still used in many applications. Phenolic resins are produced by reacting phenol and formaldehyde; the characteristics of the resin product depending on the proportions of the reactants and the catalyst employed.
Polyester Resins Typical resins, first developed in 1942, they consist of unsaturated linear polyesters dissolved in styrene. Polyester resins are rather inexpensive and have low viscosities, which is advantageous in many fabrication processes. However, shrinkage of 3-4% on curing is relatively higher than that of others.
Epoxy Resins They are comparatively more expensive and more viscous than polyester resins causing impregnation of woven fabrics more difficult. Epoxies have two or more curing stages which are major benefits since it allows performs to be pre-impregnated with a partially cured epoxy. The shrinkage on curing is in the range from 1 to 5%.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Vinylester Resins [18] Vinylesters have several advantages over unsaturated polyesters. They provide improved toughness in the cured polymer while maintaining good thermal stability and physical properties at elevated temperatures. In general, vinylesters provide excellent resistance to strong mineral acids and bleaching solutions. Most importantly, because of the basic structure of the vinylester molecule, it is more resistant to hydrolysis and oxidation than the polyesters.
Figure 5.1. The variation of residual strength and elastic modulus at elevated temperature[19]. Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
207
5.2. Failure of Thermosetting Composites by Temperature Effects The principal material strength is not just a function of either monotonic or cyclic loading, but it is also a function of time and temperature. Reifsnider and Case [12] defined that, most materials subjected to the various conditions, including temperature, stresses, etc., exhibit time-dependent fracture or stress rupture which is mainly associated with elevated temperatures. In conjunction with stress rupture, the viscoelastic nature of thermosetting polymers in particular leads their properties to time-dependent as well as temperaturedependent. Consequently, modulus of elasticity progressively diminishes as a constant load is applied at elevated temperatures. The recent work of Sul and Prusty [19] clearly demonstrates these dependency of thermosetting polymers using E-glass / polyester specimens under elevated temperature as illustrated in Figure 5.1. Since the fibre and matrix have different thermal expansion coefficients, there is a strong possibility of disturbance along the different directions of the reinforcement. This phenomenon may cause micro-stress on composites, which results in micro-cracks in the absence of mechanical stresses. High thermal expansion of the fibres can cause significant distortion of the composites. During the moulding of composites such as SMC, thermal gradients across the mould can lead to differential rates of cure and the formation of in-built thermal stresses [2]. According to Hull [8], the stresses due to curing arise from a combination of resin shrinkage during the curing processes and differential thermal contraction after post-curing at an elevated temperature. This shrinkage can lead to sink marks and other undesirable surface effects on plastic products, i.e. deteriorating the performance of the material.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
5.3. Variation in Modulus of Polymer Composites with Temperature The change in stiffness of thermosetting polymers under thermal stress only is largely resulted from the molecular rearrangements. Mahieux [20] exhibited the modulus versus temperature curve of a typical polymer, illustrated in Figure 5.2. Mahieux divided the curve into four distinct regions which are described in detail below.
Figure 5.2. Stiffness variation in polymer with a temperature increase[19].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
208
B.G. Prusty and J. Sul
The Glassy State (Region 1) The modulus of polymer often behaves steadily at very low temperature as a function of the typically order of 3 GPa. The polymer matrix experiences several transitions as the temperature applied increases. Typically the first transition is called γ relaxation, the second is termed the β relaxation, and the third is referred to as the glass transition ( Tg ) or the α transition. The γ
and β transitions (secondary transitions) reflect molecular motions
occurring in the glassy state (below Tg ). In the glassy region, the thermal energy is much smaller than the potential energy barriers to large-scale segmental motion and translation, and large segments are not free to jump from one-lattice site to another. Secondary relaxations result from localised motions. The secondary relaxations can be of 2 types: side group motion or the motion of few main chains.
The Glass Transition (Region 2) The glass transition region 2 is characterised by a steep drop in the polymer instantaneous or storage modulus. While only 1-4 chain atoms are involved in motions below the glass transition temperature, some 10-50 chain atoms attain sufficient thermal energy to move in a coordinate manner in the glass transition”. From mechanical analysis, Tg is given by the peak of the loss tangent or the inflexion point in the modulus versus temperature resulting from quasi-static experiments.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The Rubbery State (Region 3) At higher temperature (just above glass transition) a plateau can be observed. This plateau corresponds to the long-range rubber elasticity. The plateau typically indicates a modulus equal to 3 MPa. The length of the plateau increases with increasing molecular weight. The end of this plateau is characterised by the presence of a mixed region: the modulus drop becomes more pronounced but not as steep as in the liquid flow region. Short times are characterised by the inability of the entanglements to relax (rubbery behaviour) while long times allow coordinate movements of the molecular chains (liquid flow behaviour).
The Liquid Flow Region (Region 4) For linear polymers, very high temperatures can cause translations of whole polymer molecules between entanglements. The thermal energy becomes high enough to overcome local chain interactions and to promote molecular flow. Ultimately, the polymer becomes a viscous liquid and the modulus of the material drops dramatically.
5.4. Glass Transition Temperature, Tg As described above, most thermosetting polymers have a certain temperature point above which a dramatic degradation in their properties occurs. The strength of composite materials
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
209
is strongly dependent on temperature, strength and modulus rapidly decreasing once the temperature exceeds the glass transition temperature. A reduction of 20% in axial stiffness is brought about, once temperature exceeds the glass transition temperature of typical fibredominated E-glass composites [21]. This phenomenon of deterioration in stiffness is even dramatic for matrix-dominated composites, such as randomly-distributed short fibre composites. Because numerous bulk properties of the polymer undergo significant changes at the Tg , the latter value has a myriad of applications. For instance, to a component designer the Tg of a polymer represents the upper limit of its service temperature for the maintenance of
its modulus and dimensional stability [2].
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 5.3. Effect of iso-thermal heating on the mechanical property of a laminate [23].
The evolution of mechanical properties at elevated temperature presented in Figure 5.3 is not symmetrically balanced with the glass transition temperature ( Tg ) as the central point. In most fibre-reinforced composite systems the mechanical properties have deteriorated already by almost 50% prior to the Tg . This phenomenon is an important consideration in developing an analytical model estimating the mechanical properties under the temperature variation. It is desirable to relate the shape of the property vs. temperature curve represented in Figure 5.3 to the underlying distribution of relaxation times for the laminate. This requires assumptions to be made about this distribution and about the time-temperature equivalence of the material. Unfortunately, because of the complexity of the relationship between the relaxation time distribution and the property variation it is not generally possible to implement an analytical model. Most rigorous approaches therefore involve numerically fitted distributions. As far as ‘empirical’ relationships are concerned it appears that many types of polynomial functions can give an approximate fit to the data [22]. Several functions have been proposed to relate property to temperature for polymer laminates, but they fail to accurately describe the full profile of the relationship [14, 23]. The value of glass transition temperature also indicates the maturity of cure in thermosetting composites as Tg increases with the increasing extent of cure. The glass transition temperature of the polymer that have been subjected to curing process may increase compared to that of not cured polymer. Moreover, it is important to note
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
210
B.G. Prusty and J. Sul
that different thermal analysis techniques can lead to the different Tg values for the reinforcement and matrix system.
6. Fatigue Damage Modelling in Short-Fibre Composite
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The prediction of the residual life, such as residual strength and stiffness, for inhomogeneous fibre-reinforced composites is intricate. The conventional prediction methodology used for the isotropic and homogeneous materials can not be beneficial to the fatigue life prediction of fibre-reinforced composites. As discussed earlier, principal strength is the most effective mechanical property, which can be acquired from experimental characterisation, to estimate the remaining life of materials. However, such an experiment is generally burdensome, time-consuming and costly. The damage model, as a consequence of the experimental difficulties, is introduced in order to build a bridge between the gaps, for which the empirical data are not available. In spite of the fact that there is no simple justification to classify the analytical damage modelling methods in fibre-reinforced composites, many investigators [14, 24-27] classified the methodology into three broad classes, i.e. micromechanical approach (based on physical reality), phenomenological approach (based on strength or /and stiffness degradation) and statistical approach (largely based on S-N curves). Krajcinovic [26] identified each approach with a concise explanation that the micromechanical models provide fundamental information on the essential structure of the governing equations defining the thermodynamic state of the material and the kinetics of its change. Statistical methods examine the validity of certain assumptions introduced into the micromechanical models to enhance their tractability. These two classes of models ultimately provide necessary guidelines for the formulation of phenomenological models to be used in practice.
6.1. Micromechanical Model The micromechanical philosophy of Reifsnider and Case [12] defines the failure mode of composites under mechanical or/and thermal loadings which may cause fatal combinations of fatigue, stress rupture, creep and buckling (in compression). A failure function form is selected by the authors to describe the final failure event, and all of the processes that cause changes in the stress state or material state in that critical element are characterised by rates as a function of the applied conditions and generalised time.
Critical Element & Damage Accumulation Concept Case et al. [28] introduced the life prediction tool at elevated temperatures on the basis of micromechanical concept. The concept is with the assumption that the damage associated with property degradation is distributed widely within the composite laminate. They also assumed that a representative volume can be chosen such that the state of stress in that volume is typical of all other volumes in the laminate. The details of stress distribution and damage accumulation in that volume are sufficient to describe the final failure resulting from a specific failure mode. It is, therefore, required to select different
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
211
representative volume elements for different failure modes. They divided the representative volume into “critical” and “sub-critical” elements. The critical elements are chosen as their failure controls the failure of the representative volume and therefore of the laminated component. Reifsnider et al. [29] argued that residual strength may be used as a damage metric for measuring damage accumulation based on a micro-kinetic approach. Case et al. [28] then assumed that the remaining strength may be determined as a function of load level and some form of generalised time. For a given load level, a particular fraction of life corresponds to a certain reduction in remaining strength. A particular fraction of life at a second load level is equivalent to the first if and only if it gives the same reduction in remaining strength, as illustrated in Figure 6.1 showing that time t1 at an applied stress level S a1 is equivalent to time t20 at stress level S a2 because it gives the same residual strength. In addition, the remaining life at the second load level is given by the amount of generalised time required to reduce the remaining strength to the applied load level. The next step in the analysis is to postulate that normalised remaining strength (the damage metric) is an internal state variable for a damage material system. A single quantity, known as failure function, Fa , can be taken into account instead of the individual components of the strength tensor. Authors [28] constructed a second state variable, the continuity function which is defined to be (1 − Fa)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
and denoted by ψ . Residual strength is defined in terms of ψ . In order to settle thepreceding theory, it should be assumed that the kinetics are defined by a specific damage accumulation process for a particular failure mode and assign different rate equations to each of the processes that may be present. Reifsnider et al. [10] extended the preceding ‘critical element concept’ to the estimation of remaining strength and life of composite materials under mechanical, thermal, and environmental applied conditions causing the combination of fatigue, stress rupture and creep. They introduced the damage mechanics ‘continuity’, Ψ , defines in the usual way with a value of 1 when the state of the material is ‘intact’, and 0 when the material is ‘fractured’, which were adopted from [30]. The state of the material is represented by interpreting the continuity parameter as the normalised probability of survival of the material as
⎡ e ⎤ Ψ = A* exp ⎢ − ⎥ ⎣ eav ⎦
(6.1)
where e is the occupied energy level of the damage states. It was assumed that the occupied energy is proportional to the total time over which energy is supplied to the system as Ψ = A* exp ⎡⎣ −ητ j ⎤⎦
(6.2)
where τ = t / τ in which t is a time variable, j is a material parameter, and τ is a characteristic time associated with the process. Characteristic times of damage processes can be a stress rupture life, a fatigue life, a stress corrosion life, etc. Taking the natural logarithm
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
212
B.G. Prusty and J. Sul
and the variation of τ , the rate equation for the change in material state is obtained due to damage accumulation as a function of generalised time,
δΨ = −ηΨjτ j −1 δτ
(6.3)
Figure 6.1. Use of remaining strength as a damage metric [28].
Furthermore, Reifsnider et al. [10] also postulated that the continuity of the material can be set equal to (1 − Fa) where Fa is the ‘failure function’ which is a function of local stress
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
components divided by the corresponding material strength components so that they present ⎛ σ ij Ψ = 1 − Fa ⎜ ⎜X ⎝ ij
⎞ ⎟⎟ ⎠
(6.4)
As per the previous assumption that the continuity parameter is 1 with the undamaged state and 0 with completely damaged state, the following was introduced with the residual strength, Fr , as ⎛ σ ij (τ ) ⎞ Fr = 1 − ΔFa ⎜ ⎜ X (τ ) ⎟⎟ ⎝ ij ⎠
(6.5)
Then combining Eqs. (6.3) to (6.5) while taking η to be unit value, the remaining strength of the material system is obtained in the form of ⎞ σ ij (τ ) τ1 ⎛ Fr = 1 − ∫ ⎜ 1 − Fa ( ) jτ j −1 ⎟ dτ 0 ⎜ ⎟ X ij (τ ) ⎝ ⎠
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(6.6)
Fatigue and Fracture of Short Fibre Composites …
213
Reifsnider et al. [10] introduced refinement to Eq. (6.6) as fatigue is the dominant process for the determination of residual strength and assume that the characteristic ‘time’ to failure can be represented by N , the fatigue life. It is argued that life can be expressed in terms of the stress in the direction of the fibres of the most heavily loaded piles, and write a one dimensional equation of the form,
σ f (τ ) = A + B (log( N (τ ))) p X t (τ )
(6.7)
where σ f is the fibre direction stress, X is the unidirectional tensile strength in the direction
A, B, and p are material constants. In addition, they introduced the frequency of cyclic loading, f , in order to include the number of cycles of loading, n , given of the fibres, and
by n = f × t , and the following is consequently obtained as Fr (t1 ) = 1 − ∫
t1
0
⎛ σ ij (t ) ⎞ ⎛ ft ⎞ (1 − Fa ⎜ j ⎜ X (t ) ⎟⎟ ⎜⎝ N (t ) ⎟⎠ ⎝ ij ⎠
j −1
⎛ ft ⎞ )d ⎜ ⎟ ⎝ N (t ) ⎠
(6.8)
which includes the effects of fatigue, creep, stress rupture, temperature and micro-damage.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
6.2. Phenomenological Model The micromechanical approach is generally derived on the basis of simplified assumptions that strain and stress are uniformly distributed all along the constituents. The prediction of the material properties may be adequate for longitudinal properties. However, despite the advantage of micromechanical approach that can model physical reality with a minimum ambiguity and arbitrariness, it is computationally inefficient for the practical applications [26]. Since more than one of various damage mechanisms aforementioned are usually involved at the same time and interactively associated with each other, it is theoretically very difficult to construct a mechanistic model including all the damage modes. As a consequence, the phenomenological aspect has frequently been used in order to simplify the analysis of composites [24]. They also stated that from the phenomenological point of view, the damage in composites can be evaluated by the changes in material properties. On the macroscopic scale, the residual strength and stiffness is a measure for the phenomenological approach.
Combined Phenomenological Damage Model Sul et al. [14] suggested a compound fatigue damage model combining the strength and the stiffness degradation models along with temperature effect as a function of a strength variation. Ye [31] established a fatigue prediction model capable of correlating damage states, stiffness and fatigue life using a damage variable as a function of a stiffness reduction as
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
214
B.G. Prusty and J. Sul
D = 1 − ( E / E0 ) , where E is the current stiffness and E0 is the initial stiffness of the intact material. A damage accumulation law for composites can be defined as ⎛σ 2 ⎞ dD = C ⎜ max ⎟ dN ⎝ D ⎠
n
(6.9)
where C and n are material constants that can be determined by testing specimens at various stress levels or by taking logarithm at Eq. 6.9 as 2 ⎛ σ max ⎞ ⎛ dD ⎞ log ⎜ ⎟ + log C ⎟ = n ⋅ log ⎜ ⎝ dN ⎠ ⎝ D ⎠
(6.10)
On utilisation of the damage parameter and stiffness reduction in the damage variable to the damage accumulation law in Eq. 6.10, and integrating the predicted stiffness after N cycles, the estimated modulus can be expressed as E N = ⎡1 − { N ⋅ C ⋅ ( n + 1)} ⎣
1/ ( n +1)
⋅ σ max 2 n / ( n +1) ⎤ ⋅ E0 ⎦
(6.11)
Caprino and D’Amore [32], on the other hand, proposed a hypothesis that the strength of material undergoes a continuous decay under cyclic loadings as a function of the number of cycles and represented it using a power law,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
dσ N = −a0 ⋅ Δσ ⋅ N − b dN
(6.12)
where σ N is the residual material strength after N cycles, and a0 and b are material constants whilst Δσ is stress range. On integration of Eq. 6.12 to obtain σ N as
σ N = − a0 ⋅ σ max ⋅ (1 − R )
N 1− b + constant 1− b
(6.13)
As σ 0 is the strength of the virgin material, the constants are obtained by the condition
N = 1 → σ N = σ 0 . Rearranging Eq. 6.13 to include the strength degradation due to fatigue cycling can be presented in the following expression as σ 0 − σ N = α ⋅ σ max ⋅ (1 − R ) ⋅ ( N β − 1) with α =
a0 and β = 1 − b . 1− b
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(6.14)
Fatigue and Fracture of Short Fibre Composites …
215
Further, Eq. 6.14 can also be expressed in terms of σ max as
σ max =
σ0 −σ N
α ⋅ (1 − R ) ⋅ ( N β − 1)
(6.15)
To consider the temperature effects into the analytical model presented above, Sul et al. [14] assumed that the temperature is the only variable of the residual strength. Hence, a polynomial expression of 2nd order can fit the experimental data to include the temperature effect, fT as
σ N = fT (T ) = c0 + c1T + c2T 2
(6.16)
where the polynomial coefficient c with subscripts 0 , 1 and 2 can be obtained from experiments. Substitution of both Eq. 6.15 and 6.16 into Eq. 6.11 yields the proposed stiffness degradation model to evaluate the modulus after certain number of cycles as 2 n ( n +1) ⎧ ⎫ ⎡σ − c − c T − c T 2 ⎤ 1/( n +1) ⎪ ⎪ 0 0 1 2 ⎥ EN = ⎨1 − ⎣⎡ N ⋅ C ⋅ ( n + 1) ⎦⎤ ⋅⎢ ⎬ ⋅ E0 β ⎢⎣ α ⋅ (1 − R ) ⋅ ( N − 1) ⎦⎥ ⎪⎩ ⎪⎭
(6.17)
where C , n , α , β , c0 , c1 and c2 are the material parameters which are determined from
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
experimental data. The proposed model is a function of several key variables which includes the effect of fatigue cycle ( N ), temperature ( T ), stress ratio ( R ) and the tensile strength (σ 0 ) in predicting the modulus as Er ( N ) =
E(N ) = f ( N , T , R, σ 0 ) E (0)
(6.18)
7. Experimental Consideration and Verification Study The effects of fatigue loadings and aggressive environment, on the structures, often puts the maintenance engineers and designers to have a confidence regarding the fatigue life of the product or component to guarantee in-service life. Nonetheless, fatigue behaviour of fibrereinforced composites is too complex to estimate using theory alone because of their inhomogeneity and anisotropy, especially in short fibre composites [19]. Thus, the wellplanned and well-designed experiments are required due to the long time scale and the high cost of fatigue testing, it is important to choose the fatigue test conditions correctly and ensure that all test artefacts are removed or minimised [33]. There are various types and different stress ratio conditions as described in the ISO standard, ISO 13003 [34] as follows.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
216
B.G. Prusty and J. Sul
Key 1.Compression-compression region 3.Tension-tension region 5.Zero-compression alternating cycle 7.Fully reversed or fully alternating cycle 9.Alternating cycle 11.Tension-tension cycle
2.Tension-compression region 4.compression-compression cycle 6.Compression-dominated alternating cycle 8.Tension-dominated alternating cycle 10.Zero-tension cycle
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 7.1. Example of cycle types and their sine waveform [34].
7.1. Experimental Variables The dynamic fatigue test requiring a wide range of key variables that the experimentalist must consider which significantly contribute in various ways to the fatigue process. The following concise list is compiled by Andrews [21]: 1. A periodically varying stress system having a characteristic stress amplitude, 1 ⎡ ⎤ σ a ⎢σ a = (σ max − σ min )⎥ ; 2 ⎣ ⎦ 2. A corresponding fluctuating strain amplitude, ε a ; 3. A mean stress level, σ m ; 4. A mean strain,
εm ;
5. A stress ratio, σ min / σ max ; 6. A strain ratio, ε min / ε max ;
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
217
7. A frequency, ν d ; 8. A characteristic wave-form (sinusoidal, square, etc.) for both the stress and strain; 9. The ambient and internal temperature of the specimen which in general will not be the same; 10. Environmental effects; and 11. The specimen geometry
7.1.1. Stress Ratio, R Stress ratio (R) is one of the key factors that can seriously affect the result of the fatigue testing on the fibre-reinforced composites. As shown in Figure 7.1, the cyclic loading type can be determined depending on the extent of the stress ratio applied. It has been commonly reported that the lifetime of fibrous composites continuously increase as the minimum stress approaches to the maximum stress, except some cases close to the static fatigue condition. Thus, the cyclic failure time approaches the static failure as R approaches 1.0; small cyclic stresses on top of high static stresses have little effect on lifetime [35]. Under uniaxial loading, most fatigue tests from the previous investigators [36, 37] used stress ratio ranging from 1.0 (constant tension-tension) to -1.0 (tension-compression load). The S-N trend line for R = 10 show an initial drop to about 40-45% of the static strength at
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
1000 cycles, with no further strength reduction to 106 [38]. The lifetime of short fibre composites appears to be dominated by the maximum tensile stress for the stress ratios between −1.0 and +1.0 . Figure 7.2 clearly demonstrates the stress ratio effect on Glass/Polyester ( ±45º , v f = 0.38 ) showing the stress ratio of -1 apparently leads to a low profile fatigue trend compared to that of R = 0.1 . For stress ratios between −1.0 and +1.0 , the lifetimes appear to be dominated by the maximum tensile stress. Failure generally occurs in tension, with the exception of one chopped-strand-mat system at low cycles [39]. Mandell [9] summarised the effects of the stress ratio on the S-N behaviour of Chopped-Strand-Mat, namely, (1) Static fatigue ( R = 1.0 ) results for the material show a slope of approximately 6-7% of the short-time strength per decade of time under load. This is about twice as steep as for materials dominated by continuous glass fibres. (2) Since cyclic fatigue data are frequency insensitive, comparisons between cyclic and static fatigue data depend on the frequency. However, normalized S-N results at R = 0.1 are about twice as steep as R = 1.0 data when plotted vs. time to failure rather than cycles. (3) Cyclic lifetimes at R = −1.0 were indistinguishable from tensile fatigue ( R = 0 to 0.1 ) at the same maximum tensile stress for CSM. Thus fatigue data in tensiontension cycling appear to represent the approximate behaviour over a broad range of R values.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
218
B.G. Prusty and J. Sul
Figure 7.2. Fatigue behaviour of (±45º )2 s Glass-Polyester specimen [40].
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
7.1.2. Loading Frequency As a general consideration in the fatigue testing of polymer matrix composites, the proper loading frequency should be chosen to attenuate the heat caused by hysteresis in the resin and interface, and maintained constantly all along the fatigue testing. Generally, laminates dominated by continuous fibres in the test direction show lower strains and little hysteresis heating and test frequencies around 10 Hz are suitable. On the other hand, resin dominated laminates, such as CSM, show larger strains and marked hysteresis heating and frequencies of 5 Hz or less are recommended [41]. Rotem [42] studied the load frequency effect of isotropic laminates using five different load frequencies, 0.1, 1, 2.8, 10 and 28 Hz. It was found that the fatigue life decreases considerably as the frequency rises from 2.8 Hz to 10 Hz, while the changes as frequency increases from 0.1 Hz to 2.8 Hz and from 10 Hz to 28 Hz are more moderate. The effect of fatigue loading rate or the frequency on the properties is imperceptible for most continuous fibre composites. They are stressed in the fibre direction as the effect of hysteresis heating is negligible. GFRP, however, is significantly affected by the fatigue loading rate; the greater the rate of testing, greater the strength. A rate sensitivity for strength of over 100 MPa per decade rate has been reported due to the environmental sensitivity of the glass fibres rather than visco-elastic effects [41]. Furthermore, failure in a simple cyclic fatigue test would occur when the residual strength at the load rate determined by the frequency of the fatigue test was reduced to the maximum cyclic stress. Despite the inherent time sensitivity in short-fibre composites, S − N failure data for chopped strand materials show little influence of test frequency. The experimental data in tension-tension fatigue of SMC R50 in Figure 7.3 appear to be frequency insensitive beyond low cycles for a range of 2-20 Hz. Having considered to the above, the experimentalists and experimental planners are required to carry out all fatigue tests at the constant loading frequency.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
219
Figure 7.3. Tensile fatigue data at various load frequencies for SMC R50 (R=0 to 0.1) [9].
7.2. Experimentation and Verification Study
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Since the success of engineering structural materials in the market is largely costdependent, E-glass fibres embedded in thermosetting resins are widely used in practice and chosen for the verification study. Of the thermoset for the verification study is vinylester resin, widely applied in the maritime industry due to exceptional resistance to moisture absorption and hydrolytic aggression.
7.2.1. Experimental Program An experimental program was designed to perform the tension-tension fatigue test on specimens made of Chopped-Strand-Mat E-glass / Vinylester so as to verify the theoretical models. The CSM of six mats were stacked up and vinylester resins were infused between the reinforcement sheets together with a premix hardener. The fundamental properties of the fibres are 76 GPa and 3450 MPa in tensile modulus and strength, respectively while those of the cured resin are 3380 MPa and 83 MPa as shown in the table 7.1. The specimens were then cured for 24 hours and bonded with tabs, in accordance with type 3 specimen in ISO527-4 as illustrated in Figure 7.4, made of the glass fabric/epoxy in a ±45º orientation as recommended in the literature [43], in order to avoid the stress concentration at the grips. The initial average tensile properties measured are modulus of 11.25 GPa and strength of 185 MPa at 25 ºC. Uniaxial tension tests were performed on the specimens with a stress ratio of 0.1 and the loading frequency of 3 Hz using an INSTRON hydraulic universal tensile test machine. Each specimen was fatigued with a sinusoidal function and three different numbers of cycles, namely 1000, 2500 and 5000 cycles. In addition to the various cycle conditions, the elevated temperatures of 50 º C and 75 º C are used along with 25 º C in order to investigate the effects
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
220
B.G. Prusty and J. Sul
of the elevated temperatures on the thermosetting composite. Once the specimens have been fatigued, the quasi-static test was carried out on the impaired specimens until breakage at the pulling speed of 2 mm/min to define the residual properties of the specimens. The bi-axial strain gauges were bonded on the specimen in order to measure strain in the transverse direction as well as in the longitudinal direction. Table 7.1. Specification of the constituents Properties
E-Glass Fibre
Vinylester SPV 6037
Density, kg / m
2560
1040
Tensile Modulus, GPa Tensile Strength (Yield), MPa Tensile Failure Strain, % Diameter, μ m
76 108 1.8 17 6770 204
3.38 75.8 5 98.9 3100 110
3
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Heat Distortion temperature, º C Flexural Modulus, MPa Flexural Strength (Yield), MPa
Dimensions in millimetres L3
Overall length
≥ 250
L2
Distance between end tabs
150 ± 1
b1
Width
25 ± 0.5 or 50 ± 0.5
h
L0
Thickness Gauge length (recommended for extensometers)
2 to 10 50 ± 1
L LT
Initial distance between grips Length of end tabs
136 (nominal) ≥ 50
hT
Thickness of end tabs
1 to 3
Figure 7.4. Schematic diagram of the type 3 specimen [44].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
221
7.2.2. Experimental Result and Verification of the Models Residual Strength Evaluation of the empirical parameters in the strength degradation model by Caprino and D’Amore [32] was carried out. The experimental results with low cycle fatigue at the test temperatures of 25 º C, 50 º C and 75 º C are shown in Figure 7.5-7.7. The results demonstrate that the E-glass / vinylester laminate is highly sensitive to the elevated temperatures. Furthermore the residual strength deteriorates notably with an increase in temperature and subsequent decreases in strength are observed with the increase in temperatures up to 75 º C. 195
185
Residual Strength (MPa)
175
165
155
145
135
125
115 0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
Number of Cycles
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 7.5. The transition of residual strength of the specimen in fatigue at 25 º C.
A three-degradation-stage to failure of glass based composite is commonly observed in fatigue [11, 14], of which (a) a dramatic reduction stage, (b) a stable and gradual reduction stange and (c) a failure stage. On the other hand, the present results of residual strength do not appear to be the distinctive three stage of fatigue behaviour. In the investigation by Sul et al. [14], in spite of the similar experimental program except that polyester was used as matrix instead, clear three stages were observed in the transition of residual strength. The reason for the difference is that vinylester has superior fatigue resistance to that of polyester matrix in the previous study by Sul et al. [14]. Alternatively, cyclic loadings of 5000 cycles are not enough for the threshold of second stage of vinylester to fail. Moreover, it is clearly observed that the difference in temperature caused degradation in residual strength by 10 to 20 MPa. As discussed in [19], the experimental data of Chopped-Strand E-glass / vinylester laminate under fatigue are scattered with a maximum standard deviation of 12. The experimental constants of residual degradation and temperature model (Eqs. 6.15 and 6.16) are extrapolated using the least squares fit method and shown in Table 7.2.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
222
B.G. Prusty and J. Sul 195
185
Residual Strength (MPa)
175
165
155
145
135
125
115 0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
Number of Cycles
Figure 7.6. The transition of residual strength of the specimen in fatigue at 50 º C. 195
185
Residual Strength (MPa)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
175
165
155
145
135
125
115 0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
Number of Cycles
Figure 7.7. The transition of residual strength of the specimen in fatigue at 75 º C.
Table 7.2. Experimental parameters for the residual strength model and the temperature model Temperature applied, º C 25 50 75
c0
c1
c2
6.72 ×10−1
Number of Cycle 0
166.1
− 0.401
0.000271
−5
1.2
1000
187.9
−0.5196
0.000594
−25
6.51
2500 5000
191.2
− 1.137
0.007247
166.1
− 0.401
0.000271
α
β
1.44 ×10−3 1.04 ×10 1.45 × 10
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
223
Residual Stiffness The phenomenological damage model of Sul et al. [14] was examined and compared with the experimental results for fatigue loading with different test temperatures. The experimental parameters, C and n presented in Table 7.3 were obtained by the procedure similar to that in the calculation of residual strength from the low cycle fatigue experiment at elevated temperatures. The phenomenological model, Eq. 6.17 was evaluated and the residual stiffness predicted are compared with the experimental values in Figure 7.8-10. Table 7.3. Experimental parameters for the residual stiffness model Temperatu re applied, ºC 25
50
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
75
Number of Cycle 1000 2500 5000 1000 2500 5000 1000 2500 5000
Actual residual stiffness, MPa 10067.96 9837.60 9611.21 8987.24 10470.65 9676.32 9555.44 9930.43 10543.83
C
n
5.89 ×10−24
3.94
1.10 ×10−4
−7.39 ×10−1
−4.97 ×10−5
0
Predicted residual stiffness, MPa
Deviation, %
10042.10 9861.87 9615.95 9152.57 11054.88 10016.66 9049.50 9692.45 10323.64
0.26 0.25 0.05 1.84 5.58 3.52 5.29 2.40 2.09
Further analysis of the experimental data using the fatigue model of Sul et al. [14] shows a good agreement with the residual stiffness obtained by the experiment with the maximum deviation of 5.58 %. With the increase in temperature, the non-linear regression was unable to converge the mechanical properties, resulting in a larger deviation in the prediction. The fatigue model tends to overestimate the residual stiffness at 50 º C while it underestimates the residual stiffness of CSM E-glass / vinylester at 75 º C. This is due to the fact that polymer materials have remarkably low heat distortion temperature in the range of 80 º C to 150 º C, although E-glass reinforcement retains its properties up to 250 º C [14]. This characteristic of polymer resin can seriously affect the behaviour of composites under elevated temperature and make the mechanical parameter diverged. Furthermore, the stiffness increased at very low cycles as fatigue loadings applied at elevated temperatures. It is referred to as ‘wear-in’ and this has been delineated by several investigators [45-47]. The deviation can be attributed to the reasons below: (1) The redistribution and relaxation of the stresses around the notch area are the responsible mechanisms [46]. (2) The resins of thermosetting polymers are converted into hard brittle solids by chemical cross-linking resulting from thermal stress and external loads with the cross-linking leads to the formation of tightly bound three-dimensional network of polymer chains [8].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
224
B.G. Prusty and J. Sul
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 7.8. Evolution of elastic modulus of the specimen in fatigue 25 º C and its comparison with predicted data using the phenomenological model.
Figure 7.9. Evolution of elastic modulus of the specimen in fatigue at 50 º C and its comparison with predicted data using the phenomenological model.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue and Fracture of Short Fibre Composites …
225
Figure 7.10. Evolution of elastic modulus of the specimen in fatigue at 75 º C and its comparison with predicted data using the phenomenological model.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
8. Conclusion Short fibre composites are relatively inexpensive and more applicable to the massproduction than the high performance and expensive continuous fibre reinforced composites. General concepts of fatigue and fracture in composite materials have been presented in this chapter. Due to the complicated fracture and fatigue behaviour of short fibre composites, the damage model of short fibre composite is not as common as that of continuous unidirectional composite materials. Of various methodologies to predict the fatigue damage, the outline of two different approaches to prediction of fibre-reinforced composites has been presented together with the verification study of the phenomenological model. In the application of critical element methodology, the state of stress and the material should be obtained in advance [48]. On the other hand, the phenomenological model only requires mechanical parameters based on the empirical data, namely requiring a prior low cycle experiment. The mechanical properties of CSM E-glass / vinylester obtained from the verification experiment were explicitly scattering as previously reported by many investigators [14, 19, 49]. The trend of continuous reduction in residual strength was observed in the experiment under fatigue at elevated temperatures due to the process known as ‘wear-out’ whilst residual stiffness showed a queer behaviour, which is formerly disclosed [14], as the temperature increased. A recurring process on the highly cross-linked vinylester is expected to be attributed to this phenomenon at elevated temperatures. From the experimental verification, the agreement between the measured and predicted data is fairly good except for the part in which residual stiffness soared.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
226
B.G. Prusty and J. Sul
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
References [1] Williams, J.G., Fracture mechanics of composite failure. Journal of Mechanical Engineering Science, 1990. 204(4): p. 209-218. [2] Scheirs, J., Compositional and Failure Analysis of Polymers: A PRACTICAL APPROACH. 2000, West Sussex, England: John Wiley & Sons, Ltd. [3] De Paiva, J.M.F., S. Mayer, and M.C. Rezende, Evaluation of mechanical properties of four different carbon/epoxy composites used in aeronautical field. Materials Research, 2005. 8(1): p. 91-97. [4] Chawla, K.K., Fatigue, in Internaltional Encyclopedia of Composites, S.M. Lee, Editor. 1990: VCH, New York. p. 107. [5] Krenchel, H., Fibre reinforcement: theoretical and practical investigations of the elasticity and strength of fibre-reinforced materials. 1964, Copenhagen Akademisk Forlag. [6] Matthews, F.L. and R.D. Rawlings, Short fibre composites. Composite materials: Engineering and science. 1999, Boca Raton, USA: CRC Press LLC. [7] Cox, H.L., The elasticity and strength of paper and other fibrous materials. British Journal of Applied Physics, 1952(3): p. 72. [8] Hull, D., An Introduction to Composite Materials. 1981, Cambridge: Cambridge University Press. [9] Mandell, J.F., ed. Fatigue Behavior of Short Fibre Composite Materials. Fatigue of Composite Materials, ed. K.L. Reifsnider. Vol. 4. 1991, Elsevier Science Publishers: Amsterdam, The Netherlands. 231-337. [10] Reifsnider, K., S. Case, and J. Duthoit, The mechanics of composite strength evolution. Composites Science and Technology, 2000. 60(12-13): p. 2539-2546. [11] Harris, B., ed. A histroical review of the fatigue behaviour of fibre-reinforced plastics. Fatigue in composites : science and technology of the fatigue response of fibrereinforced plastics, ed. B. Harris. 2003, CRC Press: Boca Raton FL, USA. [12] Reifsnider, K. and S. Case, eds. Micromechanical models. Fatigue in composites : science and technology of the fatigue response of fibre-reinforced plastics, ed. B. Harris. 2003, Woodhead Publishing Limited: Boca Raton, USA. [13] Reifsnider, K.L. and S.W. Case, Damage Tolerance and Durability of Material Systems. 2002, New York: John Wiley Sons, Inc. [14] Sul, J., B.G. Prusty, and J.W. Pan, A fatigue life prediction model for Chopped Strand Mat GRP at elevated temperatures. Fatigue & Fracture of Engineering Materials & Structures, 2010, DOI: 10.1111/j.1460-2695.2010.01460.x. [15] Weeton, J.W., Engineer's guide to composite materials. 1987, Metals Park, Ohio : American Society for Metals. [16] Reifsnider, K. and M. Pastor, Measured Response: State Variables for Composite Materials, in Recent Advances in Experimental Mechanics. 2004. p. 87-98. [17] Matthews, F.L. and R.D. Rawlings, Polymer matrix composites. Composite materials: Engineering and science. 1999, Boca Raton, USA: CRC Press LLC. [18] Schweitzer, P.A., Corrosion of polymers and elastomers. 2nd ed. 2007, York, Pennsylvania, USA: CRC Press. 592. [19] Sul, J. and G. Prusty, Investigation on the Fatigue Life Modelling of CSM-GRP Laminates at Elevated Temperatures. World Journal of Engineering, 2009. 6(3): p. 68-74.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Fatigue and Fracture of Short Fibre Composites …
227
[20] Mahieux, C.A., A Sytematic Stiffness-Temperature Model for Polymers and Applications to the Prediction of Composite Behavior, in Materials Engineering and Science. 1999, Virginia Tech: Blacksburg. [21] Andrews, E.H., ed. Testing of polymers IV. ed. W. Brown. 1969, Interscience: New York. [22] Gibson, A.G., et al., Laminate Theory Analysis of Composites under Load in Fire. Journal of Composite Materials, 2006. 40(7): p. 639-658. [23] Mouritz, A., et al., Mechanical Property Degradation of Naval Composite Materials. Fire Technology, 2009. [24] Cheng, H.-C. and F.-S. Hwu, Fatigue reliability analysis of composites based on residual strength. Advanced Composite Materials, 2006. 15: p. 385-402. [25] Degrieck, J. and W. Van Paepegem, Fatigue damage modeling of fibre-reinforced composite materials: Review. Applied Mechanics Reviews, 2001. 54(4): p. 279-300. [26] Krajcinovic, D., M. Basista, and D. Sumarac, eds. Basic principles. Damage Mechanics of Composite Materials, ed. R. Talreja. 1994, Elsevier Science B.V.: Amsterdam, The Netherlands. [27] Van Paepegem, W. and J. Degrieck, A new coupled approach of residual stiffness and strength for fatigue of fibre-reinforced composites. International Journal of Fatigue, 2002. 24(7): p. 747-762. [28] Case, S., N. Iyengar, and K. Reifsnider, eds. Life Prediction Tool for Ceramic Matrix Composites at Elevated Temperatures. Composite Materials: Fatigue and Fracture, ed. R.B. Bucinell. Vol. Seventh. 1998, American Society for Testing and Materials. 165-178. [29] Reifsnider, K.L., et al., eds. Damage Tolerance and Durability of Fibrous Material Systems: A Micro-Kinetic Approach. Durability Analysis of Structural Composite Systems, ed. A.H. Cardon. 1996, CRC Press: Rotterdam. [30] Kachanov, L., Introduction to Continuum Damage Mechanics (Mechanics of Elastic Stability). 1986, Boston: Springer. 148. [31] Ye, L., On fatigue damage accumulation and material degradation in composite materials. Composites Science and Technology, 1989. 36(4): p. 339-350. [32] Caprino, G. and A. D'Amore, Flexural fatigue behaviour of random continuous-fibrereinforced thermoplastic composites. Composites Science and Technology, 1998. 58(6): p. 957-965. [33] Sims, G.D., ed. Fatigue test methods, problems and standards. Fatigue in Composites, ed. B. Harris. 2003, Woodhead Publishing Limited: Boca Raton, USA. [34] ISO-13003:2003, Fibre-reinforced plastics - Determination of fatigue properties under cyclic loading conditions. 2003, ISO copyright office: Switzerland. [35] Mandell, J.F. and U. Meier, eds. Effects of Stress Ratio, Frequency, and Loading Time on the Tensile Fatigue of Glass-Reinforced Epoxy. Long-term Behavior of Composites, ASTM STP 813, ed. T.K. O'Brien. 1983, American Society for Testing and Materials: Philadelphia. 55-77. [36] El Kadi, H. and F. Ellyin, Effect of stress ratio on the fatigue of unidirectional glass fibre/epoxy composite laminae. Composites, 1994. 25(10): p. 917-924. [37] Epaarachchi, J.A. and P.D. Clausen, An empirical model for fatigue behavior prediction of glass fibre-reinforced plastic composites for various stress ratios and test frequencies. Composites Part A: Applied Science and Manufacturing, 2003. 34(4): p. 313-326. [38] Riegner, D.A. and J.C. Hsu. Fatigue Considerations for FRP Composites. in SAE Fatigue Conference. 1982. Detroit, MI: Society of Automotive Engineers.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
228
B.G. Prusty and J. Sul
[39] Owen, M.J., ed. Static and Fatigue Strength of Glass Chopped Strand Mat/Polyester Resin Laminates. Short Fiber Reinforced Composite Materials, ASTM STP 772, ed. B.A. Sanders. 1982, American Society for Testing and Materials. 64-84. [40] Mandell, J.F. and D.D. Samborsky, DOE/MSU Composite Material Fatigue Database: Test Methods, Materials, and Analysis. 1997, Sandia National Laboratories. [41] Matthews, F.L. and R.D. Rawlings, Fatigue and environmental effects. Composite materials: Engineering and Science. 1999, Boca Raton, USA: CRC Press LLC. [42] Rotem, A., Load frequency effect on the fatigue strength of isotropic laminates. Composites Science and Technology, 1993. 46(2): p. 129-138. [43] Adams, D.F., L.A. Carlsson, and R.B. Pipes, Test Specimen Preparation, Strain, and Deformation Measurement Devices, and Testing Machines. Experimental Characterization of Advanced Composite Materials, 3rd ed. 2002, New York: CRC Press. [44] ISO-527-4:1997, Plastics - Determination of tensile properties, in Test conditions for isotropic and orthotropic fibre-reinforced plastic composites. 1997, ISO copyright office: Switzerland. [45] Lagace, P.A. and S.C. Nolet, eds. Effect of ply thickness on longitudinal splitting and delamination in graphite/epoxy under compressive cyclic load. Composite Materials: Fatigue and Fracture, ed. H.T. Hahn. 1986, American Society for Testing and Materials: Dallas, TX, USA. 335-360. [46] Shokrieh, M.M. and L.B. Lessard, eds. Fatigue under multiaxial stress systems. Fatigue in composites : science and technology of the fatigue response of fibre-reinforced plastics, ed. B. Harris. 2003, CRC Press: Boca Raton FL, USA. [47] Stinchcomb, W.W. and C.E. Bakis, eds. Fatigue Behavior of Composite Laminates. Fatigue of Composite Materials, ed. K.L. Reifsnider. 1990, Elsevier Science Publishers B. V.: AE Amsterdam, The Netherlands. [48] Reifsnider, K.L. and W.W. Stinchcomb, eds. A Critical-Element Model of the Residual Strength and Life of Fatigue Loaded Composite Coupons. Composite Materials: Fatigue and Fracture, ASTM STP 907, ed. H.T. Hahn. 1986, American Society for Testing and Materials: Philadelphia. 293-313. [49] Wilkinson, S.B. and J.R. White, Thermosetting short fibre reinforced composites. Short fibre-polymer composites, ed. S.K. De and J.R. White. 1996, Great Yarmouth, England: Woodhead Publishing Limited.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
In: Composite Materials in Engineering Structures Editor: Jennifer M. Davis, pp. 229-252
ISBN: 978-1-61728-857-9 © 2011 Nova Science Publishers, Inc.
Chapter 5
FATIGUE OF POLYMER MATRIX COMPOSITES AT ELEVATED TEMPERATURES - A REVIEW John Montesano, Zouheir Fawaz, Kamran Behdinan and Cheung Poon Department of Aerospace Engineering, Ryerson University, Toronto, ON, Canada
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Abstract In recent years, advanced composite materials have been frequently selected for aerospace applications due to their light weight and high strength. Polymer matrix composite (PMC) materials have also been increasingly considered for use in elevated temperature applications, such as supersonic vehicle airframes and propulsion system components. A new generation of high glass-transition temperature polymers has enabled this development to materialize. Clearly, there is a requirement to better understand the mechanical behaviour of this class of composite materials in order to achieve widespread acceptance in practical applications. More specifically, an improved understanding of the behaviour of PMC materials when subjected to elevated temperature cyclic loading is warranted. This chapter contains a comprehensive review of the experimental and numerical studies conducted on various PMC materials subjected to elevated temperature fatigue loading. Experimental investigations typically focus on observing damage phenomenon and time-dependent material behaviour exhibited during elevated temperature testing, whereas insufficient fatigue test data is found in the literature. This is mainly due to the long-term high temperature limitations of most conventional PMC materials and of the experimental equipment. Moreover, it has been found that few fatigue models have been developed that are suitable for damage progression simulations of PMC materials during elevated temperature fatigue loading. Although this review is not exhaustive, the noteworthy results and trends of the most important studies are presented, as well as their apparent shortcomings. Lastly, recommendations for future studies are addressed and the focus of current research efforts is outlined.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
230
John Montesano, Zouheir Fawaz, Kamran Behdinan et al.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
1. Introduction Considerable progress in the development of composite materials during the past few decades has enabled their widespread utilization for various industrial and recreational applications. In recent years advanced composites have emerged as indispensable materials in the aerospace industry, and as a consequence are more frequently employed due to their high strength-to-weight ratios when compared to conventional metallic components. This is exemplified by considering modern commercial aircraft such as the Airbus A380 and the Boeing 787, both of which utilize composite materials for primary structural load bearing components. The airframe of the A380 that is currently in service is comprised of more than 20% composite materials [1], mainly located in the wingbox interior structure and the aft fuselage section. Once completed, the 787 airframe will have a gross weight that is comprised of approximately 50% composite materials, which results in an aircraft that is 80% composite by volume [2]. The established acceptance of these materials in the aircraft industry and their importance for the future development of more efficient aircraft is apparent. The integration of composite materials into the propulsion systems of modern commercial and military aircraft has not experienced the same advancement. This is mainly due to the demanding temperature regime that engine components must withstand during standard operation. Nevertheless, there has been some success in using composite materials to manufacture engine components. In the mid-1990’s GE successfully integrated polymer matrix composite (PMC) fan blades on the GE90 turbofan engine. Currently GE is developing the next generation turbofan engine GEnx, which comprises of composite fan blades and an entire fan casing manufactured from a braided carbon-fiber PMC material [3]. Since these components are in the cold section of the engine, the ambient operating temperature is typically less-than 100°C. In addition, ceramic-matrix composites (CMC) and metal-matrix composites (MMC) are currently being considered as materials for jet engine components due to their superior heat resistance capabilities. Pratt and Whitney have considered a CMC material for the seals on the exhaust nozzle of the F100 PW 229 military turbine engine, while GE are considering a CMC material for the turbine vanes in the F136 developmental engine. These components are in the hot section of the engine, which can reach temperatures well in excess of 500°C. Clearly, PMC materials would not withstand long-term exposure to this severe temperature environment. There are however current demands in the industry to manufacture various structural components from composite materials for employment in the moderate temperature regions of jet engines [4], and for next-generation supersonic aircraft fuselage structures [5]. These applications demand long-term exposure to operating temperatures in the 150 - 350°C range. Fiber-reinforced PMC materials with high temperature resins may be suitable candidates for these applications, which will provide weight-saving advantages over conventional metallic components and a reduction in manufacturing costs when compared to MMC and CMC components. A new generation of high glass-transition (Tg) temperature polymers has enabled the current development of high temperature PMC materials. Consequently, high temperature PMC’s have been the focus of numerous research efforts over the past 2 decades. Both experimental and numerical studies have attempted to predict and understand the mechanical behaviour and the durability of PMC’s at elevated temperatures. More specifically, few fatigue studies on these advanced materials have been presented in the literature.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue of Polymer Matrix Composites at Elevated Temperatures - A Review
231
Understanding the fatigue behaviour of advanced composite materials is crucial for predicting their fatigue life and durability. Since aircraft components may be required to survive for over 20 years in service, the accuracy of fatigue life prediction is necessary to ensure the safe-life of composite components.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
2. Fatigue Behaviour of PMC Materials Continuous unidirectional, woven or braided fiber-reinforced PMC laminates are commonly used in critical aircraft structural parts. These materials are inhomogeneous and anisotropic, and as such exhibit markedly different behaviour than homogeneous and isotropic materials such as metallic alloys. It is therefore difficult to predict the fatigue properties of these composite materials. In general, the fatigue behaviour of metallic alloys is well understood and rather predictable. Components made from metals typically exhibit fatigue micro-crack initiation at high stress concentration locations. The gradual growth of these micro-cracks progresses for most of the components lifetime, having little influence on the macroscopic properties of the material. During the final stage, the cracks coalesce to form a larger crack which leads to rapid final failure. Once the visible dominant crack is formed, after a certain number of load cycles, the fatigue life can be determined as long as the initial crack size and its growth behaviour are known. For metallic alloys the macroscopic material properties such as stiffness and strength are unaffected or only slightly affected during fatigue loading, thus simple linearly elastic fracture mechanics models are often adopted to simulate fatigue crack propagation. Composite components on the other hand exhibit widespread damage throughout the structure without any explicit stress concentrations. Damage can also exist on both microscopic and macroscopic size scales. The common forms of damage (i.e., damage mechanisms) caused by cyclic loading are matrix cracking, fiber fracture, fiber-matrix interface debonding and delamination between adjacent plies [6]. The interaction of these damage mechanisms has been experimentally observed to have a significant influence on the fatigue behaviour [7]. Also since damage commences after only a few loading cycles and progresses upon further cycling, there is typically a gradual stiffness loss in the damaged areas of the material which leads to a continuous redistribution of stress during cyclic loading. As a consequence, simple fracture mechanics-based models are not suitable for composites since the aforementioned damage mechanisms are quite complex and the relationship between stress and strain is no longer linear. In addition, some types of composites such as cross-ply laminates have been found during cyclic loading to reach a state of damage equilibrium, which is deemed a characteristic damage state (CDS) [8]. The progression of matrix cracks in the cross-plies was found to arrest at ply interfaces and at fiber locations, causing the degradation of stiffness to vanish. Therefore, accurate prediction of composite component fatigue behaviour and fatigue life is a complex task. Experimental characterization of composites is also difficult due to the challenges in inspecting the aforementioned forms of damage and in measuring the continuous degradation of macroscopic material properties. Factors such as the constituent material properties, the fiber structure (unidirectional, woven, or braided), the laminate stacking sequence, the environmental conditions and the loading conditions (maximum stress, loading frequency) among others influence the fatigue behaviour of composites. This results in laborious and
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
232
John Montesano, Zouheir Fawaz, Kamran Behdinan et al.
costly experimental programs to characterize the material and to generate sufficient fatigue life data. This further limits the approval of newly developed prediction models since validation of a robust model must be done using various experimental test results. High temperature exposure during cyclic mechanical loading undoubtedly augments the material behaviour and the progression of the aforementioned damage mechanisms due to the potentially complex thermo-mechanical interactions. Additional property degradation mechanisms such as physical and chemical aging may continuously alter the composite properties with time, specifically impacting the polymer matrix behaviour. The influence of the time-dependent material behaviour on the fatigue damage mechanisms will also be significant at severe operating temperatures. This may in fact be the case at temperatures well below the Tg of the polymer matrix [9]. Development of a comprehensive prediction methodology for fatigue behaviour or fatigue life prediction at elevated temperatures is consequently an even more difficult task. Moreover, long-term fatigue testing at elevated temperatures poses additional difficulties due to the severe test environment, which may limit utilization of conventional fatigue testing equipment and techniques. As indicated, the continued development of high temperature polymers and their respective composites has enabled this state-of-the-art research to persist on these advanced materials.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
3. Development of High Temperature Polymers For a number of decades now, many researchers have considered the effects of elevated temperature exposure on various polymers. In the 1970’s and 1980’s, a number of high temperature polymer resins were developed by NASA as part of a larger research effort and considered as potential candidates for fiber-reinforced composite materials. Through this research, two groups of polymers known as linear polyimides and addition aromatic polyimides were developed [10]. The linear polyimides are attractive since they are both tough (i.e., high damage tolerability) and have remarkable thermal stability over a wide temperature range. The addition aromatic polyimides are more brittle, but have highly crosslinked molecular structure [11], which is beneficial for higher temperature stability where linear polyimides may fail. The main setback with these types of polyimides is that they contain known carcinogenic by-products and are very hazardous, which poses many manufacturing difficulties and risks. The first group of elevated temperature polyimide resins widely produced by NASA for use in fiber-reinforced composites was developed using a polymerization of monomer reactants (PMR) approach [12]. These addition-type polyimides were developed to have excellent thermal stability, ease of manufacturability and the ability to withstand temperatures in excess of 300°C (i.e., a trade-off between linear and addition aromatic polyimides). The static strength of these polyimides over long-term high temperature exposure was found to be fairly stable. The main derivative of this group of polyimide resins to be employed for high temperature aerospace applications is PMR-15. Many studies were conducted by NASA to improve the manufacturability and mechanical performance of PMR-15, and to ‘tailor-make’ this polyimide resin for use in fiber-reinforced composites [13]-[15]. Experimental studies were later conducted on fiber-reinforced PMR-15 composites [4]. The static mechanical property degradation, weight-loss, coupon dimensional changes, and surface thermal oxidation effects due to long-term aging at elevated temperatures were all considered during testing. Aging
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue of Polymer Matrix Composites at Elevated Temperatures - A Review
233
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
temperatures were limited to 350°C. Surface thermal oxidation was believed to be a significant contributor to material property degradation for aging greater-than 100 hours, causing microvoids and microcracks to initiate just below the damaged material surface layer. Although testing has been conducted at higher temperatures for PMR-15 composites, the maximum useful long-term operating temperature is approximately 260°C for jet engine applications. Due to the successful development and wide regard of PMR-15, additional polyimide resins were subsequently developed for high temperature composite applications. NASA also developed AMB-21 [16] and DMBZ-15 [17] high temperature polymers. These thermoset polyimides have similar properties and temperature capabilities as PMR-15, but without the hazardous carcinogenic compounds. In fact DMBZ-15 has a higher wear resistance and a slightly higher Tg when compared to PMR-15, which makes it suitable for long-term exposure at temperatures >300°C. Moreover, Dupont developed a thermoplastic polyimide Avimid K3B, which has been considered for supersonic transport aircraft [18]. The continuous maximum operating temperature for K3B is approximately 180°C. A number of additional thermoset and thermoplastic polyimide resins such as R1-16, PETI-5 and PIXA have also been considered for PMC components on supersonic aircraft with the same temperature limitations [5]. Finally, a number of high temperature BMI polymers have been developed and used in the industry. Common BMI polymers include 5250 and 5260 developed by Cytec Engineered Materials, as well as F655-2 developed by Hexcel Corporation. These polymers have a continuous maximum operating temperature of approximately 150°C. Although a number of high temperature polymers and their respective fiber-reinforced composites have been developed, there has been little use of these materials in high temperature load bearing applications. Additionally as already indicated, few studies have been conducted that consider the fatigue behaviour of these PMC materials at elevated temperatures.
4. Review of Elevated Temperature Fatigue Studies This review aims to chronologically delineate the most important accomplished fatigue studies on high temperature PMC materials. First, a discussion of the high temperature experimental work conducted on these advanced materials will be presented. This is followed by a presentation of the subsequently conducted numerical studies.
4.1. Experimental Most experimental studies focus on temperature-dependent material property degradation during static loading or isothermal aging test conditions. There are few experimental studies that consider fatigue loading at elevated temperatures. These studies will be presented, and the focus of the discussion will be on the indicated observable effects of time and temperature on the fatigue behaviour and corresponding damage mechanisms. The discussion will also include detail of the experimental test protocol and test equipment for specific studies. Lo et al [19] developed a fiber-reinforced composite which was manufactured with CSPI, a modified polyimide developed at the Chung Shan Institute of Science and Technology, having a Tg of 511°C. Isothermal mechanical fatigue testing of carbon fiber/CSPI and carbon
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
234
John Montesano, Zouheir Fawaz, Kamran Behdinan et al.
fiber/PMR-15 composites at 450°C was conducted using unidirectional and [0/90/±45] laminates. For a peak fatigue stress level of 60% of the ultimate strength and a loading frequency of 2 Hz, it was found that the fatigue life at room temperature (RT) for the CSPI composite was 106 cycles while at 450°C the fatigue life was of the order 104 cycles. The material did show a drastic decrease in fatigue life at elevated temperatures, which is not surprising. The CSPI composite demonstrated superior static and fatigue properties at elevated temperatures when compared to the PMR-15 composite. There was however no consideration for tracking the progression of damage or for quantifying material property degradation of the CSPI specimens. In addition, little information is available in the open literature to suggest any application of this material in the industry to date. Branco et al [20] conducted isothermal fatigue tests at various temperatures, stress ratios and loading frequencies for a glass fiber reinforced unidirectional phenolic resin BPJ 2018L composite up to 200°C. The glass fibers had a surface treatment applied in order to protect them from acid attack. This treatment clearly had an influence on the fiber-matrix bonding characteristics, and thus the fatigue behaviour. Stiffness degradation was monitored for both notched and un-notched specimens using an extensometer. The testing temperature was found to influence the rate of modulus reduction. It was clear that the same specimen subjected to the same loading conditions but at higher temperatures exhibited a consistent stiffness loss, whereas at RT there was little stiffness loss until close to failure. The respective plots of the normalized stiffness versus the normalized number of loading cycles for the phenolic resin composite material are shown in Figure 1. Not surprisingly, fatigue life was found to decrease with increasing temperature. Also, the fatigue life increased slightly as the loading frequency increased at the same test temperature. There is a clear time-dependence in the response of the material, which depends on the rate of stress application. Matrix cracking, fiber-matrix debonding and fiber fracture were all observed in the failed specimens using a SEM post-test. Debonding between the fibers and the matrix was deemed to be the dominant damage mechanism causing fatigue failure. Branco et al [21] later conducted the same tests using composite laminates manufactured with the same phenolic matrix with various stacking sequences. It was found that the manufacturing method (i.e., hand lay-up or pultrusion) strongly influenced the fatigue life of a specimen at elevated temperatures. Uematsu et al [22] studied delamination behaviour of unidirectional fiber-reinforced PEEK thermoplastic laminates subject to isothermal fatigue loading at 200°C. Double cantilever beam specimens were used to facilitate ply delamination during fatigue. Delamination was initiated artificially using a thin film located between adjacent laminate plies. Material stiffness was shown to decrease with increasing temperature during static testing. A fracture mechanics-based analysis was used to formulate the change in energy release rate (GI) and the corresponding stress intensity factor (K). Constant load isothermal creep tests revealed that delamination growth is continuously steady; an initial spike in delamination crack size is due to matrix degradation, while fiber bridging slows the rate of crack growth until final fracture. Fatigue loading at elevated temperatures significantly increases the delamination growth rate, which is highly dependent on the loading frequency. During higher frequency loading the crack propagation rate was completely frequencydependent or cycle-dependent (i.e., da/dN is proportional to ΔK). At lower loading frequency the crack propagation rate was independent of the frequency and completely time-dependent or creep-dependent (i.e., da/dt is proportional to K). The threshold frequency between the
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue of Polymer Matrix Composites at Elevated Temperatures - A Review
235
low/high regions was found to be approximately 0.05 Hz, as shown in the plot of crack propagation rate (da/dt) versus the inverse of the loading frequency (1/v) in Figure 2. This shows that there is little interaction between creep and fatigue mechanisms, which may seem surprising. Fatigue was therefore classified as being either time-dependent or cycledependent. Sjogren and Asp [23] also conducted a similar study to determine the effects of temperature on delamination growth in prepreg fiber-reinforced epoxy laminates subject to flexural and mixed-mode bending fatigue loading at 100°C. Delamination was also initiated artificially between selected adjacent plies. The effect of temperature on the energy release rate values for delamination growth was similarly found, where critical and threshold energy release rates decreased with an increase in test temperature. Delamination was consequently deemed to be the dominant damage mechanism causing final failure.
(a)
(b)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 1. Plots of stiffness degradation at (a) RT and (b) 100°C [20].
Figure 2. Plot of crack propagation rate as a function of 1/v [22].
Gyekenyesi et al [24] conducted an experimental study on a woven fiber-reinforced AMB21 polyimide resin matrix composite. Mechanical fatigue loading at test temperatures of 255°C was conducted using a quartz lamp as a heat source, water-cooled hydraulic grips
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
236
John Montesano, Zouheir Fawaz, Kamran Behdinan et al.
for loading the specimen, and an air-cooled extensometer to measure axial strain. Although the emphasis of the study was on the experimentation methodology, some noteworthy results for the material were obtained. High temperature static tensile tests were initially conducted, and it was found that temperature had little influence on the modulus, the ultimate stress and the strain. Fatigue loading at 255°C proved to have some influence on the fatigue life (Nf) when compared to similar RT tests. It must be noted that the coupons tested showed a large variation in fatigue lives, which was stated to be the result of poor quality specimens that included a number of defects. Ratcheting of the stress-strain curve was however observed during tension-tension fatigue, and was attributed to fiber fracture, matrix cracking, chemical degradation and mass loss due to elevated temperature and viscoelastic deformation. Ratcheting was also observed during RT testing with a similar trend in the stress-strain curve. Successive stress-strain curves for various load cycles are shown in Figure 3 for elevated temperature testing, which clearly illustrates this ratcheting behaviour. Maximum strain was used as the damage metric; there was a significant initial increase in maximum strain, followed by a gradual increase, and ending in a sudden increase before failure. The stiffness of the composite was also monitored and found to decrease continuously. Miyano et al [9] studied the effects of time and temperature on the flexural behaviour of unidirectional CFRP laminates subjected to fatigue loading. Two resin materials were considered for composite manufacturing: a general purpose epoxy 25C and a high Tg polycyanate resin RS3. Static flexural testing revealed that as the temperature increased, the mode of failure changed from tensile at lower temperatures to compressive at higher temperatures. Micro-buckling of the fibers on the compression side of the flexural specimen was observed at higher temperatures due to matrix softening. This fracture process was also observed during fatigue loading. The fatigue behaviour was remarkably dependent on both temperature and loading frequency (i.e., loading rate or time of exposure), which was attributed to the dominant viscoelastic behaviour of the matrix material. Testing revealed that increasing the test time (i.e., decreasing the loading frequency) or the test temperature caused the fatigue strength to decrease. This is illustrated in the stress-cycle (S-N) plot of Figure 4 for the RS3 resin composite. The peak testing temperature for the flexural fatigue tests was 100°C, while the loading frequency varied from either 0.05 Hz or 5 Hz. The damage due to time of exposure was found to be greater than the damage due to the number of loading cycles, which may seem surprising. Miyano and co-workers [25] continued this study where they considered frequency and temperature effects on the flexural fatigue behaviour of woven CFRP laminates manufactured from a high Tg resin 3601. Similar observations were found with the woven laminates. Case et al [26] studied the behaviour of notched unidirectional-ply fiber-reinforced K3B resin laminates. The specimens were fatigue tested at an elevated test temperature of 177°C and a constant loading frequency of 10 Hz. A convection oven was used to keep the test temperature constant, while strain gages were used for strain measurements. The strain gages did not properly bond to the specimens, thus any strain data was deemed inconclusive in this study. X-ray radiographic images near the specimen notch were also taken at various cycles in order to track damage development. Note that the fatigue test was interrupted in order to access the specimen for x-ray imaging. The dominant damage mechanisms were found to be delamination and matrix cracking. Elevated temperature was found to accelerate the
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Fatigue of Polymer Matrix Composites at Elevated Temperatures - A Review
237
delamination during cycling, while matrix cracking was also more prominent at an elevated test temperature. Castelli et al [27] tested a chopped fiber-reinforced PMR-15 polyimide matrix composite subject to combined thermo-mechanical fatigue loading at a maximum test temperature of 260°C. The material was considered for a compressor stage component in a jet engine, thus a realistic thermo-mechanical test was conducted. Fatigue damage was tracked macroscopically through deformation and stiffness measurements using an aircooled extensometer. Image analysis was used to characterize the fiber distribution orientation since the location of damage was dependent on the fiber density. Optical microscopy and SEM were used to characterize local microscopic damage. A quartz lamp system was used for heating the specimens, while an MTS load cell with water-cooled hydraulic grips was used for mechanical cycling. It was found that thermo-mechanical fatigue loading did not considerably degrade the macroscopic material properties such as axial stiffness; stiffness degradation only occurred early in cycling, and was attributed to fiber straightening. Highly localized microscopic damage was however detected at fiber bundle locations including fiber-matrix interface debonding and matrix cracking after 100 hours of cycling. Creep deformation and thus strain accumulation were however found to be significant during thermo-mechanical fatigue cycling, which was further explored through a series of isothermal stress-hold tests. Time-dependent material behaviour was found to occur at temperatures well below the Tg of the polyimide. Aging did not occur in the material after 100 hours of exposure, which was monitored by tracking the value of Tg for the polyimide matrix material. Kawai et al [28] studied the off-axis behaviour of unidirectional fiber-reinforced polymer composites subject to an elevated test temperature of 100°C and tension-tension fatigue loading. Two matrix resins were considered: PEEK and a thermoplastic polyimide resin PISP. The emphasis of the study was placed on the influence of the matrix properties, the temperature and the off-axis angle on fatigue behaviour of an elementary composite ply. A temperature chamber was used along with high-temperature hydraulic grips to load the specimens at a constant frequency of 10 Hz. Failure surfaces were examined using SEM imaging. It was found that as the off axis angle increases, the fatigue strength of the ply decreases which is no surprise. It was also found that the cyclic elastic strain range (Δε = Δσ/Ex) plotted versus the number of fatigue cycles produced two distinct linear curves, one for the on-axis (0°) loaded plies and one for the off-axis loaded plies. The normalized plot for the polyimide specimens is shown in Figure 5. This illustrates that the fibers are critical for failure for on-axis loading, while the matrix and the fiber-matrix interface is critical for offaxis loading. Also the test temperature was found to have a minimal affect on the fatigue behaviour for the on-axis loaded plies, which was not the case for the off-axis loaded plies. Off-axis plies exhibited a decrease in the fatigue strength at elevated temperatures. The failure mechanisms for on-axis loading were longitudinal matrix cracking propagating to the end tabs, whereas for off-axis loading matrix cracking and fiber-matrix debonding were the dominant mechanisms in the gage section. Also, the fatigue strength was found to change for different matrix resins, which was attributed to the varying matrix ductility at elevated temperatures, and different fiber-matrix bonding strengths. Matrix ductility was observed to be enhanced at higher temperatures, but fiber-matrix bonding was weaker as found in SEM images.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
238
John Montesano, Zouheir Fawaz, Kamran Behdinan et al.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Figure 3. Strain ratcheting at elevated temperature [24].
Figure 4. Stress-cycle plot [9].
Figure 5. Cyclic strain range vs. number of cycles at 100°C [28].
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Fatigue of Polymer Matrix Composites at Elevated Temperatures - A Review
239
Counts and Johnson [29] tested the elevated temperature fatigue capabilities of two fiber-reinforced PMC laminates, which were manufactured using PETI-5 and K3B polyimides respectively. These materials were considered for various high temperature applications that required loaded mechanically fastened joints, which in addition to elevated temperature can severely degrade the fatigue performance. Consequently, the focus of the experimental study was on the bolt-bearing capability during fatigue loading at 177°C. The IM7/PETI-5 laminate proved to have superior elevated temperature bearing fatigue properties when compared to the IM7/K3B laminate. It was found that a fatigue endurance limit existed when the maximum cyclic bearing stress was 100
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Fatigue of Polymer Matrix Composites at Elevated Temperatures - A Review
247
hours). In addition, testing at elevated temperatures has shown that material creep causes increased ratcheting of the sequential stress-strain curves when compared to equivalent RT tests [24]. Note that at RT ratcheting is typically exhibited by PMC’s, but mainly caused by damage accumulation such as matrix cracking. Therefore, time-dependent viscoelastic behaviour and degradation of the matrix material properties likely cause the changes in PMC material properties and an increase in energy dissipation during cyclic loading. Also, increasing the cyclic loading frequency is found to considerably influence the material behaviour resulting in an increase in the fatigue life of the PMC as indicated in many studies. This may be due to the increase in resistance to deformation caused by the matrix molecular structure and the viscoelastic nature of the material, or in fact be due to the reduction in exposure time at higher loading frequencies. One study reported that at low loading frequencies creep seems to be prevalent having an influence on damage propagation, while at higher loading frequencies damage propagation was more dependent on the number of loading cycles [22]. Although a higher loading frequency has been found to positively influence the fatigue life of PMC’s, higher loading frequencies have been shown to cause self-generated heating during isothermal fatigue tests resulting in higher surface temperatures. Bellenger et al [53] studied the changes in surface temperature due to load frequency of a random fiber-reinforced PA66 polyimide matrix composite subject to tension-compression bending fatigue loading at RT. Specimen surface temperatures during high frequency loading were found to increase by more than 100°C, while for lower loading frequencies the surface temperature showed a more moderate increase. Note that short fiber polyimide matrix specimens having low fiber volume fractions were used for this study, which may account for this matrix dominant behaviour. This however could suggest that there may in fact be an upper threshold for loading frequency. It is also generally observed that the growth rate of specific damage mechanisms increases with increasing temperature during fatigue loading. Shimokawa et al [54] found in their study that matrix cracking facilitated thermal oxidation, which accelerated material property degradation during long-term exposure. Additionally, the damage progression process may alter at elevated temperatures when compared to fatigue testing at room temperature. As an example, fiber-matrix debonding was observed in many studies to occur early on in fatigue testing and is deemed a primary damage mechanism [27], [52]. Debonding is typically followed by cracking of the softened matrix, then delamination and fiber fracture. Conversely, damage progression during fatigue testing at RT has often shown to initiate with matrix cracking or crazing in the off-axis plies [55], [56]. Some studies report that via observable SEM images, the fracture surfaces are similar for specimens fatigue tested at room temperature and at high temperatures [30]. In general regardless of the influence on damage, increased temperature leads to a decrease in the strength of the composite material which consequently decreases the fatigue life. There are some distinctions in the experimental observations reported which can be attributed to the variations in the type of matrix material (i.e., thermoset or thermoplastic), the composite structure (i.e., unidirectional or woven ply laminates), the loading conditions (i.e., T-T, T-C or C-C; load control or strain control; uniaxial, flexure, or biaxial) and the absolute testing temperature. The mode of loading control is a key factor that will influence the observed phenomena during fatigue testing. The experimental fatigue studies presented are all based on load-control or stress-control testing, whereas strain-control fatigue testing may alter the stress-strain phenomena (i.e., relaxation in lieu of ratcheting) and the progression of
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
248
John Montesano, Zouheir Fawaz, Kamran Behdinan et al.
damage. Another key factor is the cycling scheme (i.e., T-T, C-C, T-C), which has been shown to reveal contrasting damage mechanisms and failure modes [31]. In either case, the paramount importance of the role of the matrix material on the PMC behaviour during elevated temperature fatigue testing has been demonstrated. A major shortcoming of all the experimental studies is that there was no attempt to track damage throughout the test specimen continuously and in-situ during the high temperature tests. This is critical for accurate characterization of these materials and for improving the input to the developed prediction models. An improved experimental protocol for high temperature laboratory testing is thus required, specifically continually measuring the strain and tracking damage progression without removing the test specimen from the high temperature environment. Conventional damage monitoring techniques such as x-ray radiography, ultrasound and light microscopy require removal of the test specimen from the loading grips, while other conventional methods are not suitable for elevated temperature applications. A method using a traveling microscope was proposed by Gregory and Spearing [30] to detect delamination in double cantilevered beam specimens; the double cantilevered specimen provided an obvious delamination zone. The capability of this method to detecting other forms of damage within the test specimen must be verified by additional testing.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
5.2. Prediction Modeling As shown, there have been few studies on developing prediction methodologies for PMC’s. Although further studies can be found in the literature, they are either continuations of the presented studies or have employed similar models in their work. Of the existing fatigue models that have been presented there are few mechanistic or physically-based progressive damage models, which is seen as a clear gap in the literature. The mechanistic models that have been developed are based on unrealistic assumptions, and/or do not consider all or any forms of observable damage in the simulations. Also, few of the prediction models explicitly consider the viscoelastic effects of the PMC material, adopting linear elastic models or empirical viscoelastic factors in the prediction scheme. Since the time-dependent viscoelastic matrix has a significant influence on the fatigue behaviour, explicit consideration is considered crucial. Moreover, most models rely on extensive experimental testing to extract required empirical factors for the respective formulations. This is seen as another major drawback since the cost of testing is prohibitive in today’s aircraft industry where affordability is a major concern. In addition, all models are very specific to particular laminate composites with unidirectional or woven-plies. This limits the robustness of the prediction methodologies. The use of PMC’s for elevated temperature applications such as propulsion system components and supersonic aircraft airframes will undoubtedly increase during the upcoming years due to inevitable environmental and economic demands. Very few PMC’s that have been proven to be capable of withstanding long-term exposure at temperatures in the 150 350°C range are able to withstand mechanical cyclic loading. Clearly additional studies are required in order to gain confidence in these advanced materials, and to expand their practical usage. The development of accurate and cost-effective fatigue life prediction methodologies for PMC’s requires physically-based modeling of damage evolution, as was emphasized by Talreja [57] among others. These models must account for microscopic phenomena such as
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Fatigue of Polymer Matrix Composites at Elevated Temperatures - A Review
249
damage mechanism interactions and manufacturing defects, as well as high temperature effects such as aging, thermal oxidation, matrix degradation and viscoelastic behaviour. It is also in the opinion of the authors that accurate fatigue prediction models must account for physically-based microscopic phenomena and the associated progression of damage. The essential goal is then to relate this microscopic behaviour to the observed macroscopic behaviour of the material. This allows for simulation of the complete path of damage states during cyclic loading, which is essential for appraising the intermediate state of a material or predicting the final state of the material. Although the complexity in developing a mechanistic prediction model may be somewhat high, a few insightful simplifications may be necessary to allow for its use as a practical design tool in the industry without significantly compromising accuracy.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
5.3. Current Research Although accurate prediction models are currently lacking in the open literature, a number of recent studies have been conducted that may provide useful insight on this challenging topic. A current study by the authors [58] is attempting to develop an experimental test methodology for elevated temperature fatigue testing of PMC’s by adopting fiber optic sensors for strain and damage detection. Conventional strain monitoring devices such as strain gages and extensometers have their limitations during high-strain cyclic loading. Although strain gages have high static strain ratings, they are not capable of accurately operating at larger strains for many loading cycles [26]. Extensometers have also been shown to slip during high-strain fatigue tests, causing inaccurate strain measurements [24]. Fiber optic sensors have been proven to be sufficient for high-strain cyclic loading at elevated temperatures, which has been supported by another study conducted at room temperature [59]. For damage detection, small-scale optical sensors have been used to detect various forms of damage such as matrix cracking and ply delamination [60], [61]. The multiplexing capabilities of modern high-frequency optical interrogation devices enables continuous monitoring of test specimen damage states during cyclic loading using an array of fiber optic sensors. This state-of-the-art technology shows significant promise to be employed for real-time damage detection during high temperature cyclic loading. Regarding fatigue prediction modeling, recent studies have provided some insight on the challenges in developing accurate physically-based progressive damage models. Allen & Searcy [62] proposed a multi-scale prediction model for damage in viscoelastic solids. Multiscale prediction models of composite materials have traditionally accounted for micro-scale phenomena by employing a physically-based local representation of the constituent materials and of the local damage. The analysis results from the local scale are then input into the homogenized global scale model, which is based on the concepts of continuum mechanics. The notion is that physical phenomenon occurring at the local scale can determine the macroscopic behaviour of the material in this hierarchical domain. This is particularly attractive for engineers since FE analysis models are easily incorporated into this modeling methodology. The work by Allen & Searcy [62] extended this notion for viscoelastic composite materials. Talreja [63] developed an alternate multi-scale modeling methodology known as synergistic damage mechanics (SDM), which combines the framework of continuum damage mechanics and micromechanics formulations. Although high temperature
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
250
John Montesano, Zouheir Fawaz, Kamran Behdinan et al.
fatigue simulations were not considered in these studies, the concepts of the design methodologies may in fact be adopted. With the increase in modern computational power, the development of accurate multi-scale models may lead to adequate tools for predicting high temperature fatigue behaviour of PMC’s.
Acknowledgments The authors would like to thank the Natural Sciences and Engineering Research Council (NSERC) of Canada for a CRD grant in support of this research. The authors are also indebted to the Consortium for Research and Innovation in Aerospace in Quebec (CRIAQ) for launching and sponsoring a greater research endeavour of which this review is a component. The first author greatly acknowledges additional funding in the form of a scholarship by NSERC.
References [1] [2]
[3] [4]
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
[17] [18]
Marsh, G. Reinf. Plastics, 2002, 46, 40-43. Hawk, J. “The Boeing 787 Dreamliner: More Than an Airplane”. Presented at the AIAA/AAAF Aircraft Noise and Emissions Reduction Symposium, Monterey, CA, May, 2005, 24-26. Marsh, G. Reinf. Plastics, 2006, 50, 26-29. Bowles, KJ; Tsuji, L; Kamvouris, J; Roberts, GD. NASA Technical Report, 2003, TM-211870. Shimokawa, T; Kakuta, Y; Hamaguchi, Y; Aiyama, T. J. Comp. Mater, 2008, 42, 655-679. Jen, MR; Lee, CH. Int. J. Fatigue, 1998, 20, 617-629. Harris, B. In Fatigue in Composites; Harris, B; Ed; Woodhead Publishing Ltd.: Cambridge, England, 2003, 3-35. Highsmith, AL; Reifsnider, KL. In Damage in Composite Materials, Reifsnider, KL. Ed; ASTM STP 775, ASTM International, Philadelphia, PA, 1982, 103-117. Miyano, Y; McMurray, MK; Kitade, N; Nakada, M; Mohri, M. Composites, 1995, 26, 713-717. St. Clair, AK; St. Clair, TL. NASA Technical Report, 1981, TM-83141. Meador, MAB. NASA Technical Report, 1987, TM-89838. Serafini, TT; Vannucci, RD. NASA Technical Report, 1975, TM X-71616. Vannucci, RD. NASA Technical Report, 1982, TM-82951. Vannucci, RD. NASA Technical Report, 1987, TM-88942. Vannucci, RD; Malarik, DC. NASA Technical Report, 1990, LEW-14923. Tiano, T; Hurley, W; Roylance, M; Landrau, N; Kovar, RF. “Reactive Plasticizers for Resin Transfer Molding of High Temperature PMR Composites”. Presented at the 32nd SAMPE ISTC, Boston, MA, November 5-9, 2000. Xie, W; Pan, W; Chuang, KC. Thermo. Act, 2001, 367-368, 143-153. Sacks, S; Johnson, WS. J. of Thermoplastic Comp. Mater, 1998, 11, 429-442.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Fatigue of Polymer Matrix Composites at Elevated Temperatures - A Review
251
[19] Lo, YJ; Liu, CH; Hwang, DG; Chang, JF; Chen, JC; Chen, WY; Hsu, SE. In High Temperature and Environmental Effects on Polymeric Composites; CE; Harris, TS. Gates, Ed; ASTM STP 1174, ASTM International, Philadelphia, PA, 1993, 66-77. [20] Branco, CM; Eichler, K; Ferreira, JM. Theor. App. Fract. Mech, 1994, 20, 75-84. [21] Branco, CM; Ferreira, JM; Fael, P; Richardson, MOW. Int. J. Fatigue, 1995, 18, 255-263. [22] Uematsu, Y; Kitamura, T; Ohtani, R. Comp. Sci. Tech., 1995, 53, 333-341. [23] Sjogren, A; Asp, LE. Int. J. Fatigue, 2002, 24, 179-184. [24] Gyekenyesi, AL; Castelli, MG; Ellis, JR; Burke, CS. NASA Technical Report, 1995, TM-106927. [25] Miyano, Y; Nakada, M; McMurray, MK; Muki, R. J. Comp. Mater, 1997, 31, 619-638. [26] Case, SW; Plunkett, RB; Reifsnider, KL. In High Temperature and Environmental Effects on Polymeric Composites, TS; Gates, AH; Zureick, Ed; ASTM STP, 1302, ASTM International, Philadelphia, PA, 1997, Vol. 2, 35-49. [27] Castelli, MG; Sutter, JK; Benson, D. “Thermomechanical Fatigue Durability of T65035/PMR-15 Sheet Molding Compound”. ASTM Symposium on Time-Dependent and Non-linear Effects in Polymers and Composites, Atlanta, GA, May 4-5, 1998. [28] Kawai, M; Yajima, S; Hachinohe, A; Kawase, Y. Comp. Sci. Tech., 2001, 61, 1285-1302. [29] Counts, WA; Johnson, WS. Int. J. Fatigue, 2002, 24, 197-204. [30] Gregory, JR; Spearing, SM. Comp. Part A, 2005, 36, 665-674. [31] Shimokawa, T; Kakuta, Y; Saeki, D; Kogo, Y. J. Comp. Mater, 2007, 41, 2245-2265. [32] Hashin, Z; Rotem, A. J. Comp. Mater, 1973, 7, 448-464. [33] Reifsnider, KL; Gao, Z. Int. J. Fatigue, 1991, 13, 149-156. [34] Philippidis, TP; Vassilopoulos, AP. J. Comp. Mater, 1999, 33, 1578-1599. [35] Fawaz, Z; Ellyin, F. J. Comp. Mater, 1994, 28, 1432-1451. [36] Bond, IP. Comp. Part A, 1999, 30, 961-970. [37] Hahn, HT; Kim, RY. J. Comp. Mater, 1975, 9, 297-311. [38] Hashin, Z. Comp. Sci. Tech., 1985, 23, 1-19. [39] Whitworth, HA. Comp. Struct, 2000, 48, 261-264. [40] Hwang, W; Han, KS. J. Comp. Mater, 1986, 20, 154-165. [41] Whitworth, HA. J. Comp. Mater, 1987, 21, 362-372. [42] Yang, JN; Jones, DL; Yang, SH; Meskini, A. J. Comp. Mater, 1990, 24, 753-769. [43] Reifsnider, KL. Eng. Fract. Mech, 1986, 25, 739-749. [44] Allen, DH; Harris, CE; Groves, SE. Int. J. Sol. Struct, 1987, 23, 1301-1318. [45] Shokrieh, MM; Lessard, LB. Int. J. Fatigue, 1997, 19, 201-207. [46] Shah, AR; Murthy, PLN; Chamis, CC. “Effect of Cyclic Thermo-Mechanical Loads on Fatigue Reliability in Polymer Matrix Composites”. Proceedings of the 36th AIAA/ASME/AHS/ASC Structures, Structural Dynamics and Materials Conference, New Orleans, LA, April 10-13, 1995, Paper No. AIAA-95-1358. [47] Miyano, Y; Nakada, M; Sekine, N. Comp. Part B, 2004, 35, 497-502. [48] Kawai, M; Maki, N. Int. J. Fatigue, 2006, 28, 1297-1306. [49] Mahieux, CA; Reifsnider, KL; Case, SW. App. Comp. Mater, 2001, 8, 217-234. [50] Mahieux, CA; Reifsnider, KL; Jackson, JJ. App. Comp. Mater, 2001, 8, 249-261. [51] Sun, Z; Daniel, IM; Luo, JJ. Mater. Sci. Eng. A, 2003, 361, 302-311. [52] Jen, MHR; Tseng, YC; Lin, WH. Int. J. Fatigue, 2006, 28, 901-909. [53] Bellenger, V; Tcharkhtchi, A; Castaing, P. Int. J. Fatigue, 2006, 28, 1348-1352.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
252
John Montesano, Zouheir Fawaz, Kamran Behdinan et al.
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
[54] Shimokawa, T; Katoh, H; Hamaguchi, Y; Sanbonji, S; Mizuno, H; Nakamura, H; Asagumo, R; Tamura, H. J. Comp. Mater, 2002, 36, 885-895. [55] O’Brien, TK; Reifsnider, KL. J. Comp. Mater, 1981, 15, 55-70. [56] Razvan, A; Reifsnider, KL. Theor. App. Fract. Mech, 1991, 16, 81-89. [57] Talreja, R. “Fatigue Damage Evolution in Composites - A New Way Forward in Modeling”. Proceedings of the 2nd International Conference on Fatigue of Composites, Williamsburg, VA, 4-7 June, 2000. [58] Montesano, J; Selezneva, M; Fawaz, Z; Behdinan, K; Poon, C. “Strain and Damage Monitoring of Polymer Matrix Composite Materials at Elevated Temperatures Using Fiber Optic Sensors”, Proceedings of the SAMPE Conference and Exhibition, Seattle, WA, May 17-20, 2010. [59] DeBaere, I; Luyckx, G; Voet, E; VanPaepegem, W; Degrieck, J. Opt. Lasers Eng., 2009, 47, 403-411. [60] Takeda, S; Okabe, Y; Takeda, N. Comp. Part A, 2002, 33, 971-980. [61] Yashiro, S; Okabe, Y; Takeda, N. Comp. Sci. Tech., 2007, 67, 286-295. [62] Allen, DH; Searcy, CR. J. of Mater. Sci., 2006, 41, 6510-6519. [63] Talreja, R. J. of Mater. Sci., 2006, 41, 6800-6812.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
In: Composite Materials in Engineering Structures Editor: Jennifer M. Davis, pp. 253-291
ISBN: 978-1-61728-857-9 © 2011 Nova Science Publishers, Inc.
Chapter 6
THE CLOSED FORM SOLUTIONS OF INFINITESIMAL AND FINITE DEFORMATION OF 2-D LAMINATED CURVED BEAMS OF VARIABLE CURVATURES K.C. Lin* and C.M. Hsieh Dept. of Applied Mathematics, National Chung-Hsing University, Taichung, Taiwan
Abstract
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The analytical solutions of infinitesimal deformation and finite deformation for in-plane slender laminated curved beams of variable curvatures are developed in this research. The effects of aspect ratio, thickness ratio, stacking sequence and material orthotropic ratios on the laminated curved beams or rings are presented. By introducing the variables of curvature and angle of tangent slope, the governing equations for the infinitesimal deformation analysis are expressed in terms of un-deformed configuration. All the quantities of axial force, shear force, moment, and displacements are decoupled and expressed in terms of tangent angle. The first and second moments of arc length with respect to horizontal and vertical axes of curved beams are defined as fundamental geometric properties. The analytical solutions of circular, elliptical, parabola, cantenary, cycloid, spiral curved beams under various loading are demonstrated. The circular ring under point load and distributed load is presented as well. The analytical solutions are consistent with published results. The governing equations for finite deformation analysis are expressed in terms of deformed configuration. All the quantities are formulated as functions of angle of tangent slope in deformed state. The analytical solutions of laminated circular curved beam under pure bending are presented. The results show that the circular curved beam remains as a circular curved beam during deformation.
*
E-mail address: [email protected], Tel: + 886953002008. (Corresponding author)
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
254
K.C. Lin and C.M. Hsieh
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
1. Introduction Curved beams or laminated curved beams are used as structural members in a wide variety of applications, such as sporting goods, robotic structures, springs, and reinforced stiffeners in aircraft structures. One of the advantages of using laminated material is the finite deformation character. In practical use, the deformation is not small. Some even have apparent deformation in use such as a golf club. Due to their great importance, the literature on the static and dynamic behavior of planar curved structural elements is vast. More than 500 articles on the vibration analysis of curved beam were reviewed by Markus et al. [1] and Chidamparam & Lessia [2]. However, they were usually limited to the study of isotropic curved beams; only a few papers were devoted to the laminated composite material. In addition, the closed form analytic solutions for isotropic curved beam are less limited, not to mention laminated noncircular arcs. The finite deformation of rods in space is always related to nonlinear geometric behavior. There are two approaches which are very common. One is three-dimensional rod theory. Green & Naghdi [3] treated the rod as a three-dimensional elastic body based on three-dimensional elasticity theory. The other is one-dimensional director theory. The rod was treated as a curve by Green and Naghdi [4]. Naghdi [5]and Green [6] showed the nonlinear behavior of rods in both ways. Green [7] showed some relationship between two approaches. Due to the complexity of mathematical models, most studies have to adopt some kind of simplification and numerical methods. By using a finite element method, Li [8] derived a finite deformation theory based on total Lagrangian description for 2-D and 3-D beams of zero Poisson’s ratio without all the simplifications. Some studied the finite deformation under dynamic loading. Petyt & Fleischer [9] used three finite element models to determine the radial vibration of isotropic curved beam. They showed the cubic polynomial radial and tangential displacement field which could obtain the most accurate results. Davis et al. [10] presented the constant curvature beam finite elements for in-plane vibrations. Cheng et al. [11] developed a general finite element method using the reduced integration technique to analyze the Timoshenko beam, circular arch and plate problems. They used conforming linear finite elements for both radial displacement and rotation. Krishnan & Suresh [12] used a simple cubic linear element for static and free vibration of curved beams. Raveendranath et al. [13] assumed a cubic polynomial of the radial displacement for the analysis of laminated curved beam. Some used polynomials or power series expansion to approximate the displacement field. For example, Laura et al. [14] and Rossi et al. [15] approximated tangential displacement by using a polynomial to solve in-plane vibration of cantilevered circular arc and non-circular arcs of non-uniform cross section with a tip mass. Matsunaga [16-17] applied the method of the power series expansion of continuous displacement components to solve isotropic shallow circular arches and sandwich circular arches. Tseng et al. [18-19] decomposed the arch into several subdomains. A series solution of each subdomain was formulated in terms of polynomials and then solved in-plane vibration of isotropic and laminated arches of variable curvature. Nieh et al. [20] analyzed the elliptic arches also by using the subdomains concept to develop the displacement field. Moreover, some use suitable trial functions to approximate the displacement field. For example, Romanelli & Laura [21] used trial functions of radial and tangential displacements satisfying boundary conditions to solve the fundamental frequency of non-circular elastic hinged arcs. Wang & Moore [22] used assumed
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The Closed Form Solutions of Infinitesimal and Finite Deformation …
255
displacements satisfying the clamped ends of elliptic arc to solve lowest natural extensional frequency. Qatu [23-24] solved the free vibration of shallow and deep beams by using polynomial trial functions. There are some research studies devoted to analytical solutions of curved beams. Huang & Tseng [25] analytically performed a Laplace transform to analyze in-plane transient response of a circular arch subjected to a point load. Tüfekçï & Arpaci [26] found exact solution of free in-plane vibrations of circular arches of uniform cross-section in consideration of axial extension, transverse shear and rotatory inertia effect. Skvortsov et al. [27] studied the shallow arch of an anisotropic circular arch based on Reissner plate theory. The analytical solutions of symmetric sandwich panel subjected to symmetric loading were obtained. Atanackovic [28] analyzed the finite deformation of a circular ring under uniform pressure. Brush [29] derived a finite deformation stability equation for circular ring under various pressures. He also investigated the stability of nonlinear equilibrium equations for fluid pressure loading. Timoshenko [30] showed the large deformation of an elastica. It also showed the stability of a straight beam of large deformation. However, it is observed that the above-mentioned analytical solutions are all limited on circular beam and much of them focused on the isotropic material. Lin [31] presented the general solutions of 2-D static curved beams of arbitrary variable curvatures. He chose radius of curvature and angle of tangent slope as coordinates and derived the general solutions of arbitrary static curved isotropic beams. He then applied the solution to solve the displacement field of elliptic, parabola, hyperbola, cycloid curved beams. The circular and non-circular rings were also studied. However, his work is also limited for isotropic curved beam. In this research, a procedure similar to the method by Lin [31, 32] is extended to formulate the general solutions of arbitrary symmetric or unsymmetrical laminated curved beams by choosing the radius of curvature and angle of tangent slope as parameters. The undeformed state is adapted to analyze the infinitesimal deformation and the deformed state is applied to finite deformation analysis. Extensibility of centerline is included, and the shear deformation effect is neglected. Using the parameters, the governing equations which are developed from the balance of an element of laminated curved beam under static loading will be transformed into a set of equations in terms of angle of tangent slope. All the quantities of axial force, shear force, radial and tangential displacements of laminated curved beam will lead to decouple and be expressed as harmonic functions of angle of tangent slope. To display the solutions, the first and second moments of the laminated curved beam with respect to horizontal and vertical axes are defined as fundamental geometric properties. To validate the procedure of the present study, the analytical solutions of special cases for isotropic material are verified through comparison with the published literature. The methodology is then applied to solve the closed form solutions for various laminated curved beams such as cycloid, exponential spiral, catenary, parabola, and elliptic under various loading cases including pure bending, radial load. The circular rings subjected to point and distributed load are studied as well. In these analyses, effects of aspect ratio on elliptic arc, thickness ratio, material orthotropy ratio and stacking sequence on the behavior of laminated ring will be studied in this research.
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
256
K.C. Lin and C.M. Hsieh
Figure 1. Deformation of length element from dS to ds.
2. Fundamental Equations
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
Consider a laminated curved beam of variable curvature whose axis lies on a 2-D plane. The curved beam has a rectangular cross section of width bw and thickness h . The beam is
composed of layers of orthotropic material oriented at arbitrary angle θ with respect to the longitudinal axis. Assume the laminated beam is made of elastic material such that in each layer of laminate, the stress is linear to the strain even for finite deformation of the beam. Since the strain is finite, the displacement at a point, the extension of the axis and the rotation angle of any cross section are not necessarily small. To simplify the analysis, assume cross sections do not change the shape and size in deformed state and the cross section is always orthogonal to the axis in the deformed state. To describe the curved beam on a 2-D un-deformed configuration, shown in Figure 1, the un-deformed length element dS after deformation becomes the deformed length element ds .The coordinate of a point ( X , Y ) in the un-deformed state deforms to
(x, y ) . At the un-
deformed state, the tangent slope angle at ( X , Y ) is denoted by α . At the deformed state, the tangent slope at ( x, y ) is denoted by
where
u
ξ.
The deformation at ( X , Y ) is denoted by
is the horizontal displacement, and
v
(u, v ) ,
is the vertical displacement. For any un-
deformed length element dS , there is a corresponding radius of curvature R , such that
dS = Rdα .
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
(1)
The Closed Form Solutions of Infinitesimal and Finite Deformation …
257
Here the radius of curvature R does not have to be a constant. Most well known curves can be determined by specifying the radius of curvature, such as circle, ellipse, parabola, cycloid, hyperbola, cantenary, spiral curves, etc. The un-deformed coordinates of X, Y and arc length S are defined as: α
α
α
0
0
0
X (α ) = ∫ R( w) cos wdw, Y (α ) = ∫ R(w) sin wdw, S (α ) = ∫ R( w)dw.
(2)
Here the origin is set at α = 0. For the deformation length element ds , the corresponding radius of curvature is denoted by r , i.e.
ds = rdξ . The deformed coordinate of x, y and arc length ξ
x (ξ ) = x o + ∫ r ( w) cos wdw, 0
where
(3)
s are defined as:
ξ
ξ
0
0
y (ξ ) = y o + ∫ r ( w) sin wdw, s (ξ ) = s o + ∫ r ( w) dw,
(4)
xo , yo , so denote the deformation at the original point. The deformation is then x = X + u,
y = Y + v.
(5)
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
The deformed rotation angle ϕ can be found by
ϕ = ξ −α
(6)
To derive equilibrium equations, there are two configurations: deformed state and undeformed state which can be used. The notation and sign convention of axial force N , moment M , together with shear force V , external distributed tangential force force
qα and radial
qR are shown in Figure 2. The force balance in the reference configuration can be
expressed by
dN V N dV dM + = −qα − + = −q R =V dS R , R dS , dS .
(7)
The three equations show the balance of forces along tangential direction, radial direction and moment. The equilibrium equations can be obtained by taking free body of a curved element. These equations are the same as the force balance equation in small deformation
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
258
K.C. Lin and C.M. Hsieh
[31], since the effect of finite deformation does not effect equilibrium equation in the reference configuration.
Figure 2. Forces on a curved beam element length dS.
The strain at the centroid axis is defined by
εo =
ds − dS dS ,
or
ds = (1 + ε o )dS . As in the case of in-extensional curved beam, ε o = 0 . At a distance
(8)
z
from centroid
Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.
axis, the un-deformed length element is denoted by
dS z = (R + z )dα
(9)
ds z = (r + z ) dξ .
(10)
and the deformed length element is
The strain at a distance of
z
is defined by
εz =
ds z − dS z . dS z
(11)
dϕ . dS
(12)
Eq.(11) can be simplified to
ε z = εo + z
Composite Materials in Engineering Structures, Nova Science Publishers, Incorporated, 2011. ProQuest Ebook Central,
The Closed Form Solutions of Infinitesimal and Finite Deformation … Here assume
z